<xalbo> {klama} is a selbri, but by itself it can be a bridi
<xalbo> {mi klama} is a bridi, and the selbri is {klama}.
<kribacr> The most direct comparison between a sentence in English and one in lojban is the {jufra}.
<xalbo> Point.
jufra is a sentence. What you say between full stops (or {.i}s in lojban).
For all intents and purposes, jufra are the same thing as English sentences.
bridi are what are fundamentally different.
bridi are relationships expressed within jufra.
{mi klama} is a bridi, as well as a complete jufra, or sentence.
But if you say {mi klama lo zarci ku} - that is a bridi, but {zarci} is also the selbri of another bridi.
For any given selbri, it is describing a bridi relationship.
I'm short on time still, so I can't really delve into it more deep than that. I hope I've been some help.
<Suprano> can you construct me an ambiguous example for elidable terminators?
<xalbo> I can't construct an ambiguous example, because as I said before, any text stream has at most one parse.
<xalbo> No lojban text is ambiguous, ever.
<xalbo>
What happens if you omit terminators you wanted to include is not that
you create an ambiguity, but that you get something that unambiguously
parses in a way you didn't intend.
Okay, I'll start off with the terminology update.
As
I think I mentioned before, early in my teachings I dodge as much
terminology as possible to make things easier to transition.
Now that you're feeling more like a citizen of lojbanistan, it's time to dump some more on you.
First, we'll start with ideas and concepts you've already learned, and expand the terminology of those concepts.
You know sumti, selbri, bridi, and jufra.
There are also cmavo...
{cmavo} "x1 is a structure word of grammatical class x2, with meaning/function x3 in usage (language) x4."
{jufra} "x1 (text) is a sentence/statement about x2 [topic/subject/predicate/focus] in language x3."
jufra = sentence, consisting of one or several bridi.
{.i mi djica lo nu citka lo plise}
That whole thing is a jufra.
The whole thing is also a bridi.
{citka lo plise} is also a bridi.
cmavo are the structure words of the language.
<marenz_> like se nu kei be bei be'o?
<kribacr> Yep. Those are cmavo.
Now, selbri are sort of broken up into a few categories.
Most of the selbri you've seen so far are a type of brivla called gismu.
{gismu} "x1 is a (Lojban) root word expressing relation x2 among argument roles x3, with affix(es) x4."
gismu are the five-letter rootwords.
{gismu} is itself a gismu, as are {dunda} {pendo} etc.
{brivla} "v1 is a morphologically defined predicate word signifying relation b2 in language v3."
brivla specifically have a consonant cluster within the first five letters.
Any brivla can be used as a selbri.
selbri are the relation definers.
brivla are a kind of selbri that have a consonant cluster.
gismu are a kind of brivla that have 5 letters.
<gunkamanti> cmevla are not brivla, right?
<kribacr> Correct.
<kribacr> cmene are names
{cmene} "x1 (quoted word(s)) is a/the name/title/tag of x2 to/used-by namer/name-user x3 (person)."
<kribacr> cmevla are "name words" - words like .kribacr. or .lojban.
<gunkamanti> so, everything with consonant cluster in the 1st 5 letters, and ends with a vowel, is a brivla.
<kribacr> Yes.
<kribacr> Those are the two rules for brivla.
<kribacr> Well, three rules. Penultimate stress.
<kribacr> I feel like the Spanish Inquisition all of a sudden.
The
rules for identifying a brivla are a bit more complicated than that.
For example "lomlatu" has a consonant cluster in the 1st 5 letters,
ends in a vowel and has penultimate stress, but it is not a brivla.
Neither is "slinku'i" a brivla. If you already know that you are
dealing with a (single) valid Lojban word, then you can say that if it
contains a consonant cluster and ends in a vowel it is a brivla (no
need for the "1st 5 letters" part, "mu'aupli" is a valid fu'ivla, even
though the cluster is not within the first five letters).
{fu'ivla} "v1=f1 is a loan-word with meaning v2 in language v3, based on foreign-language word f2"
{lujvo} "x1 (text) is a compound predicate word with meaning x2 and arguments x3 built from metaphor x4."
fu'ivla are borrowed words.
Words that are native to another language but are in some way lojbanized for our use.
lujvo are compounded words that are built together by means of rafsi and given a specific definiton.
You've already run into a number of lujvo.
<marenz_> rafsi?
rafsi are three, four, or five letter word-parts that you use to make lujvo.
99% of all gismu have at least 2 rafsi, and any given amount up to 5.
Note: brodi, brodo and brodu are the only gismu with only one rafsi (their 5-letter-form).
<gunkamanti> possible rafsi for fukpi are -fuk- and -fu'i- (and -fukp- and -fukpi, which are default)
<kribacr> lujvo and their construction is a very complex topic.
<kribacr> But if you see a selbri with more than 5 letters, it's probably a lujvo.
<kribacr> selbri is a lujvo
{selbri} is made from the tanru {se bridi}
{fu'ivla} is made from the tanru {fukpi valsi}
Some common lujvo are {gendra}, {xrukla}, the aforementioned grammar word ones, there's a good amount you'll see.
Some cmavo also have rafsi - {se} is probably the most commonly used one for that purpose.
{gendra} "d1 is grammatically correct in situation d2 according to grammar d3=g1."
{gendra} is made from the tanru {gerna drani}
{gerna} "x1 is the grammar/rules/defining form of language x2 for structure/text x3."
{drani} "x1 is correct/proper/right/perfect in property/aspect x2 (ka) in situation x3 by standard x4."
You need not memorize all these now, they are just mentioned to give you a general impression.
<kribacr> Trying to think what terms I missed.
<gunkamanti> for now, keep in mind that many rafsi look the same as some cmavo, but they're almost never related.
<gunkamanti> rafsi never occur alone.
If you look at gismu lists and they have letter forms next to them, those are rafsi.
<gunkamanti> it's like "forgotten", which starts with "for" but has (?) nothing to do with it.
Now, some cmavo are based off of the structure of gismu that are related to them.
But it's important to separate rafsi.
{se} has a rafsi as well - {sel}
It's the {sel} part of {selbri}.
Some cmavo also have rafsi - {se} is probably the most commonly used one
<gunkamanti>
so, in some special cases, there are cmavo that have itself as a rafsi
(eg so'i), but generally, rafsi + cmavo just look the same, are
completely different.
<xalbo> rafsi are a
different type of thing; where cmavo, brivla, and cmevla are all
*words*, rafsi are not words; they are parts of words.
<gunkamanti> {se} is a cmavo, having the rafsi {sel}
I should have said {se} is the most commonly used cmavo that has a rafsi.
<xalbo> .i zoi .raf. sel .raf. rafsi zo se
<xalbo> -sel- is a rafsi for the word {se}.
<marenz_> if rafsi are parts of words, what part of {se} has {sel}, having even more letters?
Well, they are parts of words that they construct.
They are only lexically related to the words they are based off of in best approximation.
<xalbo>
Rafsi are strings of letters, frequently mnemonically related to the
words that they rafsi. They can be combined (according to complex
rules) into words. They can *not* be used as words on their own.
Oh, I forgot gadri.
{gadri} "x1 is an article/descriptor labelling description x2 (text) in language x3 with semantics x4."
lo / la / le are gadri
<marenz_> rafsi are a certain mutation of their original word, used for combining it with others
<kribacr> Yep.
<xalbo> The words that are formed by combining rafsi are called lujvo.
You
by now covered most of the terminology. Again, I don't expect you to
remember or master all of them right now. I'll be restating each type
of word as I teach new material.
<marenz_> so, gismu and cmavo each have rafsi associated with them. What else?
<xalbo> All gismu, but only some cmavo have rafsi.
<kribacr> That's it; only gismu and cmavo.
<marenz_> So you can't make a lujvo out of every cmavo+gismu.
<kribacr> Not every cmavo. But all gismu.
Interesting aside: some lujvo translate -extremely- humorously into English or vice versa!
{staple} (rafsi sta - ple, from {stali pelji}): "staying-paper"
{stifle} (rafsi sti - fle, from {sisti flecu}): "ceasing-current(s)"
{recyclable} (rafsi recy - cla - ble, from {rectu clani ruble}): "meat-ish-long-weak thing(s)"
Note the needed y in recyclable
{je'e} "vocative: roger (ack) - negative acknowledge; used to acknowledge offers and thanks."
The part after the hyphen is the -nai version.
je'e - Acknowledge / je'e nai - Do not acknowledge
{ui} "attitudinal: happiness - unhappiness." ui - uinai
<kribacr> That reminds me, is there a positive affirmer for UI?
<xalbo> {ja'ai} is an experimental cmavo that's been proposed, because officially, no.
{au} "attitudinal: desire - indifference - reluctance." positive - neutral - negative
some even have three definitions, this one is for au, aucu'i, aunai
So far, we have learned how to play with selbri in a few ways.
We can change the place structure using {se} and friends.
From
an English perspective, we talk about cmavo by their cmavo groups, or
selma'o - and we indicate these through capital letters.
So {se} cmavo we call SE.
fa, fe, fi are FA
lo / le are LE
Each cmavo has a grammatical category to which it belongs that defines its grammar.
ui u'i je'e are UI
nu / du'u are NU
All cmavo of the same selma'o can go in the exact same spots in the grammar (obviously with different meanings)
We can kinda manipulate place structure using {be}.
But some gismu that describe relationships might lack some places that could be pertinent from time to time.
{viska} "x1 sees/views/perceives visually x2 under conditions x3."
viska lacks a place for what is being used to do the seeing. (Probably eyes, but could be something else.)
{cusku} "x1 (agent) expresses/says x2 (sedu'u/text/lu'e concept) for audience x3 via expressive medium x4."
cusku lacks a place for the language used to express the statement.
{cmavo} "x1 is a structure word of grammatical class x2, with meaning/function x3 in usage (language) x4."
selma'o: Grammatical class.
{selma'o} - x1 is the grammatical class of structure word x2, with meaning x3 in language x4.
terminators are normally in their own class; for example, ku is KU
{ku} "elidable terminator: end description, modal, or negator sumti; often elidable."
{ku}
glossing to "end sumti" of selma'o KU terminating LE, LA elidable
terminator: end description, modal, or negator sumti; often elidable.
{lo} glossing to "the really is" of selma'o LE terminated by KU veridical descriptor: the one(s) that really is(are) ...
{du'u}
glossing to "bridi abstract" of selma'o NU terminated by KEI with rafsi
-dum- abstractor: predication/bridi abstractor; x1 is predication
[bridi] expressed in sentence x2.
Anyway, selbri often lack places that could be useful in some situations.
lojban provides us with a mechanism to add places to selbri on the fly, to give more meaning to what we say.
That mechanism is {fi'o}.
Let's look at an example.
{cukta} "x1 is a book containing work x2 by author x3 for audience x4 preserved in medium x5."
{cukta} is a gismu, with five places - all of which are useful and potentially very pertinent to any given conversation.
{.i ti cukta} — This is a book.
{.i ti cukta fi mi} — This is a book written by me.
Now...
What if this book was written in lojban?
There's no place in {cukta} for the language of the book.
<marenz_> we could specify the x2 more elaborate
<kribacr> Well, that's more about the work of the book, not the language it's in.
<kribacr> A potential solution, possibly.
We could make a tanru, as well:
{.i ti lojbo cukta fi mi}
Not guarenteed to mean what you want.
But there's an easier way.
{fi'o} will allow us to add a new place to an existing selbri, and then subsequently fill that place with something meaningful.
You can think of {fi'o} as another FA-like tag - but one that you get to define.
So if {fa} marks the x1, {fe} the x2 and so on, {fi'o} marks the place of another selbri that you import.
{bangu} "x1 is a/the language/dialect used by x2 to express/communicate x3 (si'o/du'u, not quote)."
The x1 of {bangu} is a language - this would be a meaningful slot to add to the {cukta} we want to describe.
So we say...
{.i ti cukta fi mi fi'o bangu la .lojban.}
This is a slight derivation from the "x2 always starts on the right side" thing, but there's a reason for that.
So we have....
{ti (x1 of cukta)
{mi} (x3 of cukta)
{la .lojban.} (x1 of bangu, which we have made a new place for cukta)
Another example:
{.i mi cusku} — I express.
{cusku} "x1 (agent) expresses/says x2 (sedu'u/text/lu'e concept) for audience x3 via expressive medium x4."
{.i mi cusku fi do} — I express things to you.
{.i mi cusku fi do fi'o bangu la .lojban.} — I express things to you in lojban.
{.i mi viska} — I see.
{.i mi viska do} — I see you.
{pilno} "x1 uses/employs x2 [tool, apparatus, machine, agent, acting entity, material] for purpose x3."
{pilno}'s x2 is a tool of some sort.
To access that x2, we will use {se}.
{.i mi viska do fi'o se pilno lo minra} — I see you using a mirror.
{fi'o} has a slightly strange form as you might have noticed - first you say {fi'o}, then a selbri, then the x1 of that selbri.
{fi'o} does have a terminator - {fe'u} - which isn't needed too terribly often.
{fi'o} can go anywhere a FA tag can go, as it sort of does the same thing - marks a sumti.
{fe'u} might crop up if you tag the place before the selbri.
<kribacr> Or actually, no it wouldn't.
<kribacr> *thinks*
<xalbo> {fe'u} would be needed if you wanted to make an ad-hoc selbri tcita
<kribacr> I can't think of any case off of the top of my head where it's needed. I'm sure there is one.
<kribacr> Can you think of an example?
{tcita} "x1 is a label/tag of x2 showing information x3."
<xalbo> {mi fi'o lenjo fe'u viska} I see, with lenses (probably glasses) something unspecified.
<kribacr> Oh yeah, I had almost forgotten that construct.
So as you can see, {fi'o} opens many doors for even more complex sayings. There are a number of gismu that get used fairly often with the {fi'o} idea - but oddly enough, {fi'o} doesn't see that much use. (Which is a shame, because it's awesome.)
What people DO use are BAI tags.
They do the same thing, but they are shortened versions of the {fi'o [selbri] [sumti]} concept.
For example, instead of {fi'o bangu} you can just say {bau}.
Instead of {fi'o pilno} you can just say {pi'o}. (or {fi'o se pilno} {se pi'o}.
BAI tags are cmavo that are very closely tied in appearence to the gismu they represent.
<kribacr> lujvo are compound words made from rafsi.
<marenz_> fi'o bangu bau sounds like a luvjo to me
<kribacr> Like I said, lujvo are compound words created from rafsi.
{rafsi} "x1 is an affix/suffix/prefix/combining-form for word/concept x2, form/properties x3, language x4."
<kribacr> lujvo are brivla, and thus have to have 5 or more characters.
<marenz_> thats what I meant. it looks like a luvjo
<marenz_> to words that become one
<marenz_> *two
<marenz_> but i understand the difference
-> there cannot be any space between rafsi in a lujvo.
BAI
are more like rafsi, in that they are shortened forms of words, used
for a different purpose. But BAI and rafsi are distinct (although there
are some letter strings that are both BAI and rafsi derived from the
same words, just like there are some rafsi that are textually identical
to their selrafsi)
The only other 'structure' things I can think of that we haven't covered are relative phrases and logical connectives.
But
the extra place tagging we just went over is a pretty broad concept
that doesn't involve just extra sumti places. Tenses can also be
sumtcita.
The BAI tags are just shorter versions of common {fi'o} candidates.
{.i mi cusku fi do bau la .lojban.}
==
{.i mi cusku fi do fi'o bangu la .lojban.}
{.i mi viska do fi'o se pilno lo minra}
==
{.i mi viska do se pi'o lo minra}
There are about 50 BAI tags, though only a dozen or two see regular use.
http://www.lojban.org/publications/wordlists/cmavo_selmaho_order.txt
BAI tags cannot be built on the fly. They are a set list. They are simply a collection of shorthand {fi'o}-like tags. That link lists all of them.
The ones labelled BAI* are just SE derived versions of the core list.
However, you can create your own {fi'o} tags on the fly. You just can't give them the shorter form.
{fi'o} is really useful in that aspect.
People don't use it as often as they could / should because it's sort of diminshed in the texts that exist.
I teach it first because it's arguably more powerful than the BAI tags.
I suppose I should cover questions, commands, and quotes.
Commands are pretty easy to cover, since as that I know there's only one command word in all of lojban.
{do} "pro-sumti: you listener(s); identified by vocative."
{ko} "pro-sumti: you (imperative); make it true for you, the listener."
{do} is exactly the same as {ko}, with the exception that it turns the statement into a command.
As the definition seems to imply, it gives a conveyed meaning that the listener should make the statement true.
When we gloss {ko} commands, we usually just keep the same text and say "Make it true that ..." at the beginning.
{.i do citka lo fraso cirla} — You eat some French cheese.
{.i ko citka lo fraso cirla} — Make it true that you eat some French cheese.
Note
that lojban commands can be fairly interesting - they might not convey
things so directly or with the same wording that a command in English
might.
Examples:
{.i mi dunda lo sanmi do} — I give a meal to you.
{.i mi dunda lo sanmi ko} — Make it true that I give a meal to you.
Maybe you need to wash your hands or something before eating. Or sit down at the table. Or whatever.
There will be no direct assumption of hauteur, aggression, or any other emotion with a command. They are just that - a command.
"command" in itself has loads of connotations, "imperative statement" is maybe better.
I can cover quotes now.
Quotes come in ... five different forms.
Three different methods for quoting lojbanic words, and two for quoting non-lojbanic words.
The simplest example is {zo} - which quotes a single lojbanic word, namely, the next word you say.
so it doesn't have a terminator: it self-terminates.
{cusku} "x1 (agent) expresses/says x2 (sedu'u/text/lu'e concept) for audience x3 via expressive medium x4."
{.i mi cusku zo coi} — I say "Hello."
{.i mi cusku zo blanu} — I say "Blue."
As said, no terminator necessary, thus:
{.i mi zo blanu cusku} — I say "Blue."
That does not form a tanru.
{blanu} is simply quoted text, as marked by {zo}.
I put both sumti before the selbri.
{.i zo blanu se cusku mi}
Same thing.
So - {zo} quotes single words. Simple.
{selsku} "c2 is said by c1 to audience c3 via expressive medium c4."
Next is the quotes for a longer statement.
{lu}, which is terminated by {li'u}.
The text between {lu} and {li'u} must absolutely be grammatical.
But beyond that, any lojbanic text will suffice.
It can be pages upon pages upon pages as long as it is grammatical.
What if the quoted text quotes? As it is nested, it is no problem:
{.i mi cusku lu .i coi pendo .i mi'e .i mi cusku lu .i mi merko li'u li'u}
<xalbo>
Since you've been asking about the grammar, the text inside {lu} quotes
is just another call to the top level production in the grammar. So it
can embed just fine.
The last way to quote things said in lojban is {lo'u} and {le'u}.
These work just like {lu} and {li'u} but can take non-grammatical lojban.
The text still must be lojbanic words, but it does not have to be grammatical.
<marenz_> but then it it can't have {li'u}
<xalbo> It can't have {le'u}. It can have anything else.
<kribacr> {lo'u} {le'u} can. {le'u} can't be quoted by that construct, true.
<kribacr> xalbo: You could theoretically JOI with a {zo} inbetween if you had to quote a {le'u} in a big string of non-grammar stuff though, right?
<xalbo> Yes, I think {ce'o} is the standard one for that purpose.
<xalbo>
There was also at one point a proposal to use {zo le'u} as an escaped
{le'u}, but I don't think it ever really took off.
So, {zo} for one-word quotes.
{lu} {li'u} for grammatical quotes of any length.
{lo'u} {le'u} for quotes of any length. They can be grammatical, but don't have to.
As far as the terminators go, they can only be left off in really rare occasions.
"le'u"
is never elidable. Any following word would be grabbed as part of the
quote anyway, but even at the end of the utterance it is not elidable.
(I
would argue that a truly, truly fluent lojbanist could understand when
they are left off, but for now, the general rule is "just don't".)
Examples:
{mi'a xirma sonci .i mi'a cusku zo ni} — We are knights; we say "ni".
{mi'a} "pro-sumti: me/we the speaker(s)/author(s) & others unspecified, but not you, the listener."
{oi} "attitudinal: complaint - pleasure."
{lo nu do na dunda lo cmalu tricu mi cu mukti lo nu mi fi do cusku fe zo ni}
— If you dont give us a shrubbery, we will say "ni!"
{xirma} "x1 is a horse/equine/[colt/mare/stallion/pony] of species/breed x2."
{sonci} "x1 is a soldier/warrior/fighter of army x2."
{cusku} "x1 (agent) expresses/says x2 (sedu'u/text/lu'e concept) for audience x3 via expressive medium x4."
{lo'u xirma sonci cusku le'u zo'e preti do la makfa}
— You just asked makfa "xirma sonci cusku".
{la .kribacr. pu fanva lu lo'u xirma sonci cusku le'u zo'e preti do la makfa li'u}
{lu .i vrude .i to'e vrude .i mi ponse po'o lo xumjimcelxa'i li'u se cusku la .ac.}
— "Good, bad, I'm the guy with the gun." said Ash. Nov 18
Dbrockman@googlewave.com:
Rather, it means “Good. Evil. I just own a gun.”
"ba'e
mi te xumjimcelxa'i" is better for "*I'm* the one with the gun", but
even better is "ba'e mi te xarci", simply because "xarci" is a common
word in Lojban, like "gun" in English, while "xumjimcelxa'i" is an
unlikely monster.
(although kribacr's version asserted that all he does with the gun is own it, which seems a little less threatening)
{xumjimcelxa'i} "x1
= c1 is a gun/chemically launched metal slug throwing weapon for use
against x2 by x3; weapon fires metallic objects x4 using chemical
propellant x5."
