You already know the five basic predicate conversion prefixes (SE): zo, zu, za, ze, zi. However, unlike Lojban, Gua\spi some additional ones, which might take some time getting used to, but which probably will come in handy at some point. Here they are:

• zm  = (1 2 3) becomes (2 3 1)
• zr   = Switch x2 and x3
• zl    = Switch x2 and x4
  
• zy   = Switch x2 and x5
  

I'm not sure if these are going to be of much use yet, but they exist.

Unrelatedly, here the beginning of an interlinnear of the text on the whiteboard from an earlier post. Would you find such a thing helpful for the entire text? It's a lot of work to do this...

"Tigereye cries to the children: "The barge (boat carrier), look out!"

Notes:
This is the first sentence of a new paragraph (indicated by |vi gza (gza = x1 is the first sentence of a paragraph)), and so it makes use of a goi-like mechanism; we assign Tigereye the word do (the first of three ko'a-variables), by using the word xim = x1 is called/has name x2, attached with a restrictive subordinate clause.
qu is a lot like Lojban's {la}; the following phrase is turned into a name.
The word qe acts like a {ki} but it can sticky any modal case you want, here it makes it so that every following sentence is also equipped with ^ve kam, i.e. the yelling continues (and the one who does the yelling as well as the one being yelled at stay the same) until a new sticky tense is used, or the old one is removed.
There are many subordinates in this sentence, so it might be hard to follow. If you take all of them away, all that's left is "^:a jiw \su cana fer" ~ "(someone) is surprised about the barge". Of course, with all the discursives the sentence is turned into an emphatic, yelled, warning.
Pay close attention to where all the different phrases end, keep in mind that a 5th tone always ends the previous one and then starts a new phrase on the same level. With the above explanations and the previous posts you have the necessary tools to understand the structure of this sentence. As always, leave a comment if you have questions or suggestions. (I know the formatting is a bit ugly, but I don't know how to fix that, and unless someone complains, I don't intend to)

End of Part 12

My attempt to translate this famous passage to Gua\spi. There is still much room for improvement, but I really need to talk to Jim Carter to sort some questions out, and until then, this is as good as I can do.

; Morphology.
C           = (choice of letters)
Cseq     = (Cseq C) | C
V          = (choice of letters)
Vseq     = (Vseq V) | V
Word     = Cseq Vseq

; Tone categories.
Compound    = `-' | `='
Sametone     = `^'
Down1          = `!' | `|'
Up1               = `/'

; Grammar. LHS `-' symbol indicates which end has a tone.
Prefix          = (subset of Word, e.g. ``vo'' or ``zu'')
Primitive     = (subset of Word, e.g. ``tara'' or ``crw'')
Phrase         = Prefix Args0 Phrase
    | Phrase-w
Phrase-w     = Primitive Phrase-w
    | Primitive
Phrase0-      = Phrase Sametone
    | Phrase Down1 Args1

; Args(n) is a list of phrases that jumps up n levels at the end. Args3, 4, . . . are defined similar to Args1 and 2. Some finite bound must be set on n to give a finite grammar.
-Args0-     = Compound (Just one tone)
    | Down1 Args1
Args1-      = Phrase Up1
    | Phrase Down1 Args2
    | Phrase0 Args1
Args2-      = Phrase `!' `fu'
    | Phrase Down1 Args3
    | Phrase0 Args2

; These add the effect of ``fi''.
Afterargs     = Phrase0 Afterargs
    | Phrase Down1 Afterargs
    | Phrase
After1     | After1 Down1 `fi' `-' After1
    | After1 `-' `fi' Down1 After1
    | Afterargs
-Preargs-     = Down1 After1 `/fi' Preargs
    | Down1 After1 Down1 `fi' `/' Args1
    | Args0
-Sentstart     = `^' (choice of words such as ``:i'')
Sentend      = Phrase Down1 After1
    | Phrase
-Sentence     = Sentstart Preargs Sentend
-Nonsentence    = Sentstart Down1 Sentend
-Unit     = Sentence | Nonsentence
-Discourse     = (Discourse Unit) | Unit

"Discourse'' is the root grameme. Grammar for quoted non-gua\spi text is not shown, but foreign predicates and quoted gua\spi are processed by this grammar and are put together at the organizational syntax level.

Negation is relatively well documented in the reference grammar, I don't think it would be useful to restate everything. Therefore I'll just quote it here (I added the numbers to the examples and made the style fit the rest of my example sentences):

In gua\spi negation is not a unitary concept; beside the obvious antonyms there are nine or ten ways to express negative meanings, most of which involve compound words. Here are a few examples:

(1) ^:i sfa \kio \ji ^tara |zey \ju
      "It is false that I have your rat."

This is the prototype of negation, and it is the policy in gua\spi to use predicates when possible rather than prefixes or other structure words. However, the negated sentence is an extra level down, a problem for speakers.

(2) ^:i go \ji /kio \tara |zey \ju
      "I don't have your rat."

``go'' is a mood prefix which means that the asserted sentence is counter to fact. It is simpler and more familiar to natural language speakers than ``sfa - false'' is, and it works in subordinate clauses where ``sfa'' doesn't.

(3) ^:i \ji /kio \tara |go zey \ju
      "I have a rat which isn't yours."
``go'' can equally be used in subordinate clauses, or even in argument predicates.

(4) ^:i \ji /kio \xn kseo
      "I have no cheese."

``xn'' means that of the members of the full referent set of the argument, none fit in the predicated relation. Unlike the rest of the articles, this is actually a statement about the excluded members, and means the same as ``^:i \ji /go kio \xa kseo''  freely translated, ``for all pieces of cheese, I don't have it''.

(5) ^:i \ji /kio \kseo |zu cy
      "I have zero pieces of cheese."

This is the most natural form of argument negation in Loglan, but gua\spi looks strictly at referent sets, and if you say you have all the members of the null set, it isn't a cheesy null set  there is only one null set. The statement is a tautology, and says nothing about cheese. Many logical fallacies, such as St. Anselm's ontological proof of the existence of God, are like this example in that they prove a statement about the members of a set which may not have any members. In gua\spi use ``xn'' as above.

(6) ^:i \ji /kio \ple \tara
      "I have something which isn't the rat."

The full referent set of ``ple \xe tara'' (and therefore its referent subset) is in the complement of the referent subset of ``xe tara''.

(7) ^:i \jw |kseo /fi stu zao
      "This cheese is bad in flavor."

In George Orwell's 1984, the language ``newspeak'' was designed to destroy the ability of people to think, and one of its design features was that negative words were eliminated; ``bad'' became ``ungood''. Gua\spi (imitating Loglan) offers specific negated words for major predicates when the negations are used frequently. Nonetheless, most negations will have to be done with compound words as in the next examples. Be alert for creative expression possibilities such as ``^:i \jw |kseo \fu zu dyi''  ``this cheese is disgusting''.

(8) ^:i \jw |kseo \fu /gl zao
      "This cheese is flavorless."

Many dimensions are quantifiable (more or less) but unsigned, so their degree ranges from zero to larger values. This is how to assert that the degree is zero or negligible.

(9) ^:i \jw |kseo \fu /gr ksi
      "This cheese is not fresh."

When the dimension ranges from positive to negative values, ``gr'' interchanges positive and negative. On occasion, ``gl'' will also apply to indicate the zero point, though it is meaningless with ``ksi - fresh''. For extremes of unfreshness one can use ``fpu - rotten''.

(10) ^:i \jw |kseo /fi vry can psl
        "This cheese is desolidifying."

When a process occurs in the reverse of the usual order, ``vry - reverse'' indicates this.

End of quote. If you need to look up any words, use the dictionary.

So as you can see, a very elaborate system of negation is in place, we have NAhE, we have selbri-NA, we have bridi-initial NA ku, everything is there. Of course this isn't Lojban, so you won't be sliding around {naku} and change scope. Still, negation does scope over quantifiers, just as it does in Lojban.
If you have questions, leave a comment below.

End of part 11

Tenses were covered indirectly in the subordinate clause article since tenses work the same way; you import some modal case that specifies when the bridi happens. Here are the three basic tenses:
cnu = x1 is at the same time as x2 — present tense
bir = x1 happens before x2 — past tense
jro = x1 happens after x2 — future tense

You use these as modal cases, but you don't have to fill them:

(1) ^:i \ji /crw |jro
      "I will eat."

(2) ^:i \ji /cyr |jro \crw \ji
      "I go after I eat."

Just like the rest of modals, you can move them around freely. For aspects, the same is true, but they are also often used in compounds, for example tiu = X1 continuously is or does (vo) X2:+1, which is perfect for an infinitive compound:

(3) ^:i \ji /tiu pur
      "I continuously run."

Other times, you use them as a modal case ((4b) is just another way to say it):

(4a) ^:i \ji /crw |scu ^kseo
      "I completely eat the cheese."
      "The event of me eating the cheese is complete."    

(4b) ^:i \ji /crw \kseo ^vu scu
      "I completely eat the cheese."

Here is another example, using siw = X1 ceases to be (vo) X2:+1:

(5) ^:i \ji /siw pur |cnu \jai \ju ^vam \ju ^ji
      "I will stop running when you say you love me."
      .i mi co'u bajra ca lo nu do cusku fi mi fe lo se du'u do mi prami

The Gua\spi sentence in almost every example is a bit shorter than the English, and in (5) much shorter than the Lojban, thanks to the tones, subordinate mechanisms and short brivla.

If you want to have sentences in sequence e.g. "I ate. Then I ran.", you can use the sentence starter :a instead of the usual :i:

(6) ^:i \ji /crw ^:a \ji /pur
      "I ate. Then I ran."

Finally, here is a table of PU and ZAhO correspondence in Lojban and Gua\spi:

The online dictionary seems to work now (For some reason it didn't work when I last checked). Try it, it's way nicer than ctrl-F'ing through the .dat file.

http://www.math.ucla.edu/~jimc/guaspi/xankua.html

All the basics of Gua\spi grammar have been covered, but of course there are several important details that are worth exploring. One reader asked about mathematical expressions in Gua\spi, and indeed the author of Gua\spi considers MEX to be the core of Gua\pi. Being a huge fan of mekso in Lojban, I can only be excited about this. And from what I can see, Gua\spi does it a bit less awkwardly.

So let's see what Gua\spi does with numbers. First of all, here are the digits from 0 to 9:

You will note that the pattern of the vowels is the same as in the case link prefixes (FA) and the predicate conversion prefixes (SE).
I already hinted in previous articles at the fact that numbers in Gua\spi are predicates, and that's entirely true. Every number has the place structure: x1 is a set of n members x2, in extension" (where n is the number/digit) If we apply a conversion with zu, it will be "x1 is a member of set x2 which has n members".
So the word cu would mean "x1 is a set of 2 members x2, in extension".
Let's see how we could use such a number in a sentence (1) and (2).

(1) ^:i \kmau /zu jiau \zu ce
       "The cat has four legs."
       lit. "the four members of some set are the legs of the cat"

(2) ^:i \ji /vyi \xgom |zu ka
      "I see eight dogs." ("I see dogs which are eight in number")
    
Sentence (2) uses a relative clause to attach a number to an argument, this is very common in Gua\spi.

We can also ask about numbers, like {xo} in Lojban. The word pa has the place structure: "x1 is a set of how many members x2 ?". Here's an example sentence (3):

(3) ^:i \ju /crw \qkao |zu pa
     "How many cakes did you eat?"

Okay, but how do we talk about the number itself. How do we say "1 + 1 = 2" ? Gua\spi thinks about numbers as sets, and the set of all the numbers in a set is the number itself. The article xu gets the entire referent set of an argument, as a set (or class). Let's look at an example (4).
plw = x1 is the sum of x2 (xu), x3 (xu), x4 (xu)... (things in brackets always indicate a default article)

(4) ^:i \xu cu /plw \co ^co
       "2 is the sum of 1 and 1."

As you can see, we don't need to place xu before the numbers because this article is there by default in plw2-n, but not in plw1. Here's one with subtraction, using the same predicate:

(5) ^:i \ca /zu plw \xu ka ^ci
      "3 is the difference 8 minus 5"

Ordinals are constucted using the word tr, whose place structure is: x1 is nth in list (xy) x2 starting at x3:

(6) ^:i \tr cu \dwu /fi kmau
      "The second animal is a cat." (maybe several animals are standing in a line)

(7) ^:i \ji /tr cu \sty kqa \diu sui
      "I am the second smallest in my class."

sty = List (xy) x1 is a list of set (xy) x2 in order (vo) x3
kqa = x1 is big
diu = (xy) x1 is a team for activity (vo) x2+1
sui = x1 learns to do skill (xo-vo) x2 from teacher x3

It's like 40 degrees today, so I don't feel like writing more. Sentence (7) might be too difficult...

These are the basics of how Gua\spi treats numbers. "[The] syntax for mathematical expressions is neat, compact and unambiguous. No special syntax needs to be added to gua\spi beyond that already in use for ordinary arguments and sentences." — from the reference grammar.
If you have any questions, leave a comment below.

End of part 9

Gua\spi has neither tanru nor lujvo. Instead, it uses very strictly regulated compounds. There are three different kinds of compounds, depending on their components' place structures / kinds of relations. A compound always takes an even tone, either the 1st or the 6th.

Infinitive Compounds
As the name suggests, these compounds have in it a predicate that contains a default case for an infinitive, for instance daw (x1 wants to do/be x2 (vo)). In Gua\spi compounds, the tertau comes first, a bit like in Vietnamese, for example daw cyr has daw as the tertau, and cyr as seltau (we'll see later whether or not it's even useful to think of things as ter/sel-tau).
To get the place structure of such a compound, we just fill the vo-case of the tertau with the seltau, and also copy the x1 of the tertau into the x1 of the seltau. An example will help:

(1) ^:i \ji /daw cyr \dowu \ji
       "I want to go home [to my house]."

daw = x1 wants to be/do x2 (vo)
cyr = x1 goes to x2 from x3 via x4

If we want to find out the place structure of the compound daw cyr, we do this:
daw = x1 wants to be/do x2 (vo)
cyr =                                 x1 goes to x2 from x3 via x4
daw crw = x1 wants to go to x2 from x3 via x4

That was a very simple example, but it can get a *bit* more tricky. Sometimes it's not the x2 that has the default vo-case, but the x3 or some other case. Here's such a case (no pun intended):

(2) ^:i \ji /qma pue \snan ^jume

qma = x1 makes x2 do/be x3 (vo)
pue = x1 spreads over/into x2 from x3 via x4
snan = x1 is some sand
jume = x1 is a table

qma has the vo-case in the x3, which means that the following predicate word will be placed in that place and not in the x2. Here's what happens:

x1 makes x2 do/be x3 (vo)
                               x1 spreads over/into x2 from x3 via x4
x1 makes x2            spread over/into       x3 from x4 via x5

So we arrive at the following translation:

(2) ^:i \ji /qma pue \snan ^jume
      "I spread sand over the table." ("I make sand spread over the table")

This shows that sometimes some of the other arguments of the tertau get filled before the merged seltau place gets filled. It's always regular, however. Also I find that it's possible to mentally jump through the place structures of the component predicates, so that you don't even need to keep track of how many places the lujvo has, you just jump through the necessary tertau places, then the seltau places, which is not so hard because it's so predicable.
You can also make compounds of more than two predicate words:

(3) ^:i daw can fti \ji
       "I want to become healthy."
       "I want to recover."

So keep in mind this rule of thumb: infinitive + 1st tone = infinitive compound

Shared Argument Compounds
This second type of compounds is very simple. It's exactly the same as Lojban's {broda je brode}; all the arguments are shared between broda and brode. Such a compound automatically occurs when a 1st tone joins predicates whose tertau has no case for a default infinitive. So when do we want such a compound, you ask?  Here are two examples:

(4) ^:i \ji /pur far \sa tiri
      "I run away from the tiger."

(5) ^:i \ji /suy pne qmy \kqua
      "I swim down through/into the water."

In each case it's possible to state this as seperate bridi, just like in Lojban.
(4) expands to:
\ji /pur \sa tiri x2                   +     \ji far x2
"I run from the tiger to x2"           "I am far from the destination of my running"
                        "I run to far away from the tiger."
                            "I ran away from the tiger."

This is pretty clever. The unfilled places are also merged, and thus you get this very interesting way of saying this. Also, sa would fill the x3 of far, but it has no x3, so nothing happens.

(5) expands to:
\ji /suy \kqua         +    \ji pne \kqua                       +     \ji /qmy \kqua
"I swim to water"         "I penetrate that water."          "I am above the water."

All of them happen at the same time, so I am above the water as I begin to penetrate it, and I'm also swimming towards it, i.e. I'm swimming down through the water, I'm plunging into the water. So when talking about downwards movement, we are using the word "above", and the same will be true for other directions; their polarity gets reversed. You could say "I swim from above" and then it's not so weird.

Even though these shared compound predicates are technically simple, I think they will require some rethinking on my part, it's a different way of thinking about certain expressions, even compared to Lojban.
Lastly, if you want to make a shared argument compound, but the tertau has an infinitive case, then you can use fe to override the infinitive-compound default. fe is like {je}, which makes sense.

So keep in mind these rules of thumb: non-infinitive + 1st tone = shared argument compound
                                                              infinitive + fe + 1st tone = shared argument compound

Transitive Predicate Compounds
Okay, this one is the most confusing for me. This type occurs when you have a transitive predicate and use the 6th tone. By transitive predicate, they mean an agentive one, in Lojban terms. Took me some time to figure out that this was the meaning they intended by "transitive", because in logic it means something entirely different.
If you use the 6th tone in a compound, it puts the seltau into the second case of the tertau, just like in the first type of compound, but here there is no infinitive, it's just an object.
I'll show you the example from the reference grammar, because I wouldn't know a good example myself:

(6) ^:i spo \bri =kqua |bir ^dri =fli

Okay, holy shit. This has some crazy nesting going on, and it's the only example they give, and it uses the 6th tone twice and not even as a selbri in one case! Yay for easy (not!) examples.
spo = x1 (vo) is possible
bir = x1 (vo) is in the past of x2 (vo)
bri = x1 breathes x2
kqua = x1 is some water
dri = x1 drives x2 to x3 from x4 via x5  (transitive)
fli = x1 flies to x2 from x3 via x4  (non-transitive / intransitive)

I think the nesting goes like this:
^:i spo
             \
                bri = kqua
                               |
                                bir    ^   dri = fli

And to me it looks like the beer case, I mean, bir-case isn't filled at all, which means it's just a past tense marker in the middle of nowhere, okay.
Okay, "something is possible", so much is clear. Now what is bri =kqua? According to the rules, this puts kqua into bri2, so it should mean "breath water", the intended meaning is "drown". This is a selbri and not a sumti here, because spo1 is a default vo. So then, we have thus far "It is possible that someone drowns". Next up is the ominous bir-modal. There can't be anything in it because the following ^-tone ends a phrase and starts a new one on the same level. So I believe this is like {pu ku}. Maybe this is how to construct afterthought tense, or tense at all, I haven't seen any other examples of tense. Cool, interesting. Last but not least, dri =fli. Again, the 6th tone puts fli into dri2, that is "drive something that flies to x2 from ....", and here it's used as an argument of bri =kqua, so we are told who is drowning: it's the operator of the flying thing, a pilot, so to speak. Thus I arrive at the translation "Maybe the pilot drowned". The text tells me it means "Maybe the pilot already drowned". Close enough, the "already" is probably there for no reason. If you found a mistake in my explanation, please tell me! Here it is again for you:

(6) ^:i spo \bri =kqua |bir ^dri =fli
       "Maybe the pilot drowned"

What happens to the remaining places of dri? Does the lujvo have 9 places or what? Not to mention that they would all be doubled, because both times we have "to x2 from x3 via x4". Hmm.

Also, I'm wondering why you can't just use the 4th tone instead of the 6th here, as in:

(7) ^:i spo \bri \kqua \ji
       "Maybe I am a breather of ... OH!
Okay, so as you see, this puts kqua in bri1, and now I would have to use annoying FA to first put kqua into bri2 and then ji into bri1. Okay, so it's just a handy short-cut it seems.

So keep in mind this rule of thumb: transitive tertau + 6th tone  = transitive compound

I guess a native speaker of Gua\spi wouldn't even think about it, but for me, it's still not so easy to see when I want which kind of compound. I know this was a somewhat lengthy article, but I'm happy it's out of the way now. I hope this is of some help to you out there trying to get a grasp of Gua\spi.

End of Part 8

Today I gave gua\spi its own section on this blog, but that meant I had to manually copy each article over, which was rather annoying.

Gua\spi has very few selma'o, just five in fact, as opposed to like 50 in Lojban, and you can fit the entire formal grammar on a single sheet of paper (A4 paper, normal size font, for all the nit-pickers si Lojbanists reading this). How is this possible?paunai

The five categories into which you can sort any gua\spi word are:

• The word fu  (a version of {cu}, to bridge large gaps in nesting)
  
• The word fi  (a version of {cu} for smaller jumps) 
• Sentence start words (e.g. :i, :e, :o etc) 
• Other prefixes (stuff like SE and FA and subordinators)
  
• Everything else (i.e. brivla). This category is so extremely huge because even things like articles and numbers have place structures in gua\spi.
  

The word fi is rather common, but I haven't seen fu used yet. fi is like {cu}, but it's special in that the tone you use on it has a somewhat different effect than on any other word in gua\spi. There are three different tones you can use with fi, and the tone used determines how far upwards in the nesting you jump. To figure out where you will end up, just imagine that you are currently on level 2 (if 1 is the highest). Then:

• /fi will always bring you to level 1/top level
  
• ^fi will always bring you to level 2
• \fi will always bring you to level 3

And these work no matter where you actually currently are in the parse tree. The same idea applies to fu, but it jumps by steps of two (I've never seen it used). Sentence (1) is an example of fi.

(1) ^:i \ftu |crw \ji /fi ksu
      "The apple I ate was the delicious."

Sentence start words you've already seen in form of :i. The others are the same but signal some extra semantics. For example, :a means that the sentences occur in sequence. :o is much more interesting; it's a sort of {.i TAG bo}, bit only sort of. Comapre (2a) and (2b).

(2a) ^:i \vo \ji /crw /cnu \byw \ji
        "The event of me eating is at the same time as the event of me drinking."
        "I eat and drink at the same time."

(2b) ^: \ji /crw ^:u cnu \byw \ji
        "I eat, which is at the same time as my drinking."

What happens is that :o takes the previous sentence and sticks it into the first case of the predicate of the next sentence. That predicate has to accept an event in its x1 though, or else it's not smudra. Doing this saves the use of explicit infinitives/abstractions, just like {.i TAG bo} or probably {la'e di'u} does in Lojban, and it's more afterthoughty, easier on the brain for the speaker. The inventor of Gua\spi asks the learner not to rely on it or to not use it at all, however. I think it should be field-tested extensively before drawing any conclusions.
We are also already familiar with different types of prefixes, you know the case links [so, su, sa, se, si] and the predicate conversions [zo, zu, za, ze, zi]. Prefixes use the 1st/compound tone. Negators are also prefixes.
And you also know how predicates work, obviously, so that means that you now know every single selma'o gua\spi has to offer. Yay!

End of Part 7

[The following article is almost a week old, I just added a few sentences, marked in colour, and deleted those that don't fit anymore (due to me finally understanding these things). Anyway, I could have completely rewritten the article, but decided to keep it the way it was when I first had these experiences and emotions.]

Oh. My. God.

There are so many things you need to know, but I must try to explain things in order. This language is amazing. What I'm describing today was the hardest part (but also the most rewarding) of this language for me, (for a stupid reason). I can assure you though, that once you understand this part of gua\spi, not even the skai is the limit.

Subordinate clauses are very very common in gua\spi, and because of that, a special tone is given just to them. The 3rd tone. You use the word vu to signal a subordinate clause. One of the most important and common usages of subordinate clauses is to restrict an argument or to restrict the main sentence.
When the third tone is present, vu is automatically assumed and can be left out, but when the third tone does not apply, vu must be used explicitly.
There are two additional rules "to make subordinate clauses simpler":

• When the predicate of a subordinate clause has a case for an event, the predicate is automatically converted to put the event first.
  
• The restricted phrase is automatically replicated in the first case of the clause which, if the previous rule applies, will be the event argument.

So a lot happens automatically.  Let's look at an example (1a).

(1a) ^:i \tara /crw \xi kseo |vu xel
       "The rat eats smelly cheese." (xi is an article, similar to {su'o}, I'll explain articles in another article)

xel = x1 smells like x2.
|vu attaches to kseo (cheese) (one level below) and adds a restrictive clause to it. The first case of xel is automatically filled by kseo. We literally get "cheese that smells". It's common to use subordinates instead of adjectives in gua\spi. We can also omit the vu (1b), because it's implied by the 3rd tone anyway. Note the change in tone.

(1b) ^:i \tara /crw \xi kseo |xel
       "The rat eats smelly cheese."

Okay, this one I understand well enough (I now completely understand it). But now, there are several different types of subordinates, each with their own linking word:

• vu: A restrictive clause, which events of the modified phrase must satisfy, or they are thrown out of the referent set.  I think of it as {poi}
  
• ve: A supplementary comment, giving additional information about the modified phrase, typically adding a modal case.  This is the gua\spi alternative to Lojban's BAI
  
• vi: A discursive comment, a helpful assertion by the speaker of the relation between the modified phrase and the previous sentence.  A mixture of UI and SEI and some other juicy stuff
  
• va: A supplementary assertion often stating the speaker's relation to the modified phrase. Basically a {noi}
  

The one that still confuses me is the second one, ve (not anymore :) ). Even though they all work the same, the fact that they are used in different situations makes them different in a way. I know that ve is like BAI, but look at this example (2) and try to figure it out along with me.

(2) ^:i \tara /crw \kseo ^ve tum \vden \xgno
       "The rat eats the cheese with its teeth."

First of all, vocab: xgno = vo'a (gua\spi has a huge number of such pronouns), vden = x1 is a tooth of x2. tum = x1 uses x2 for purpose x3 (=pilno) (And this is where I went wrong and why I ran around so confused I thought I was going crazy or stupid because I just couldn't see how the places rearranged accordingly, but today I found out that tum = x1 is a tool for  activity x2 (vo)!).
Secondly, we can see that ve takes the 5th tone, so it's placing something on the sumti level, that is, it's subordinated to the selbri. The subordinate clause it adds to the sentence contains additional places, which are called modal cases. This is how gua\spi does BAI, so yes, there are no BAI.
So the ^ve tum \vden is like {fau lo nu tutci fa lo denci} except fau isn't exactly right. But the difference between BAI and this is that BAI only adds a single place each time, and subordinate clauses can add as many places as there are places in the subordinate clause (minus one for the restricted clause, which has to appear in it).
Let's look at another example, this time we create a tense BAI. This is the great thing about SC (sub. clauses, sick of writing it out), you can use any predicate you want and use it as a BAI or TAG or whatever you want to call it. To me this means that speech is much more conscious, and you are forced to know the gimste well, because it's all you have and need, so it probably makes you more eloquent in the process, because you learn to juggle with ad-hoc brivla tag thingies all the time. I don't know about you, but I think this is incredibly cool.
So let's use bir = x1 is before x2 (= {purci}} and use it as a modal case. Here is the same sentence in a few different variations.

(3a) ^:i \ji /crw |ve bir \vo pqi \ji
        "I eat before I sleep."

(3b) ^:i \ji /crw |bir \vo pqi \ji
        "I eat before I sleep."

(3c) ^:i \ji /crw |bir \pqi \ji
        "I eat before I sleep."

(3b) and (3c) show an advantage of the "No-empty-x1 rule"; without it, we couldn't omit the vo at all.
I suppose you could omit the second ji as in (3d).

(3d) ^:i \ji /crw |bir \pqi
        "I eat before (I) sleep."

I'd add more examples of SC, but I'm sick of this article. If you want examples, tell me in the comments and I will provide some.

Using subordinate clauses, you can do a lot. So much, in fact, that gua\spi doesn't need BAI or UI (yeah I already told you now...). That's right, there are no BAI (!)(or in other words, everything is potentially a BAI) and no UI (!) (again, everything is). BOOM. Language just got 25 times simpler.
Another thing that doesn't exist is terminators. There is no need for them because you always know when a phrase ends by hearing the tone. Remember when I said there were only two types of words? cmavo and gismu, but now I can tell you that almost all cmavo are pronouns and numbers (and even those are actually all predicates!) and the rest are function words or structure words (of which there aren't many). The vast majority of words are predicates, and maybe you can see how this is possible now. I'll talk abou selma'o in the next article.

End of part 6