Re: [lojban] what's a du'u?

[Daniel Brockman <daniel@brockman.se>]

On Thu, Dec 3, 2009 at 3:21 AM, Thomas Jack <thomasjack@gmail.com> wrote:
>
> Supposedly a du'u is a predication, a bridi.  [...]
>
> I think I must have misunderstood something along the way,
> because if {lu co'e li'u bridi} is true, and a du'u really is just
> a bridi, I'd think we could just use lu/li'u in place of du'u.

You have to be careful here, because the word "bridi" can refer to two
separate things: either a string of text, or a predication.

Now, a du'u1 is a predication, a du'u2 is a string of text, and a {lu}
expression refers to a string of text.  So while you definitely cannot
use {lu broda} in place of {lo du'u broda}, you can pretty much use it
in place of {lo se du'u broda} (although {lu broda} is more specific).

A separate but related question is whether bridi1 corresponds to du'u1
or du'u2.  The gimste pretty much explicitly says that bridi1 is a
string of text, and so corresponds to du'u2, but I think we should
redefine bridi1 to instead correspond to du'u2.  We had a discussion
about this on IRC as recently as yesterday.

Here's a slightly edited transcript:

<dbrock> I think we should make an effort to redefine bridi1
         to be a du'u1 rather than a du'u2
<xalbo> I think all of those words are about text:
        {bridi}, {sumti}, etc.
<dbrock> is {lo broda} a bridi?
<Twey> No, it's a sumti.
<Twey> It contains a bridi.
<Twey> A sumti is, by definition, only part of a bridi
<dbrock> so {broda} in {lo broda} is a bridi?
<Twey> Yes
<xalbo> The {broda} in {lo broda} is *not* a bridi.  {broda}
        can, by itself, act as a bridi, but it isn't doing
        so there.
<xalbo> Just as the word "flies" can be a verb (as in "Time
        flies like an arrow"), but the "flies" in "Fruit flies
        like a banana." is not.
<dbrock> Twey: there are four distinct entities relevant to {lo
         broda}: two strings of text, one containing the other;
         one predication; and one predicate argument
<dbrock> neither of the text strings are bridi, grammatically
<dbrock> oh, there's also a fifth entity: the predicate
<Twey> Question-begging
<dbrock> I'm just trying to be precise
<dbrock> I know I haven't addressed your assertion that {broda}
         as part of {lo broda} is a bridi
<Twey> Ah, okay
[Twey shuts up and waits.]
<dbrock> the problem is that {bridi} is ill-defined
<dbrock> that's why I reverted to English
<dbrock> xalbo's position is that bridi1 refers to a syntactic
         element, i.e., a string of text in a certain
         grammatical context
<dbrock> which can probably be characterized by being terminated
         by {vau}
<dbrock> (hi donri!)
<xalbo> I think {vau} is indeed relevant.
<xalbo> {lo broda ku} is what we're talking about, if we use the
        full structure.  No place for a {vau}.
<dbrock> Twey's position appears to be that bridi1 refers to any
         string of text that when placed between, say {.i
         ... .i}, becomes a xalbo-bridi1
<dbrock> my position is that bridi1 is not a string of text at
         all, but rather a predication
<dbrock> (but let's get back to that)
<dbrock> one problem with your definition, xalbo, is that there
         is no place for the grammatical context
<dbrock> which makes Twey's argument that {zo broda bridi}
         (regardless of context) kind of strong
<xalbo> That does appear to be a good argument.
<dbrock> but I don't like Twey's position
<Twey> My position is that ‘lo broda’ is in fact a predication
<dbrock> it seems both bad and Lojbanically counter-intuitive
<Twey> ‘lo’ just ‘raises’ the first sumti of the predication so
       that the overall effect is ‘that which is…’ instead of
       ‘something is…’
<dbrock> Twey: let's distinguish between a predication and a
         string of text that refers to a predication
<Twey> I did so
<dbrock> but `lo broda' is unarguably a string of text
<kribacr> I've come to like the term {ka'erselbri} or
          {selbrika'e}.
<Twey> dbrock: I was, of course, referring to the predication
       represented by the string of text
<dbrock> Twey: okay, but I object to the use of "of course" here
         :-)
<Twey> Well, *any* example given on IRC is going to be a string
       of text.
<dbrock> after all, to me this is _exactly_ the cause of the
         confusion
<Twey> But a string of text, of course, cannot be a predication,
       so I can't possibly have been referring to the text
       directly
<dbrock> but a string of text can be a bridi
<dbrock> and a bridi is a predication
<dbrock> ... or is it?
<dbrock> etc.
<dbrock> see my point?
<Twey> A string of text cannot *be* a bridi — it *represents* a
       bridi
<Twey> A string of text is nothing more than a bunch of
       characters in sequence
<dbrock> the gimste disagrees
<dbrock> (and you disagreed a minute ago)
<Twey> Where?
<dbrock> hmm, maybe I just misunderstood
<dbrock> 20:28 <dbrock> so {broda} in {lo broda} is a bridi?
         20:28 <Twey> Yes
<Twey> I think the gimste means to say that the text
       *represents* a predicate relationship.  Of course, nobody
       actually says that normally, since it's obvious that text
       itself cannot be anything other than a series
       of squiggles.
<xalbo> A bridi is, by the definition in the gimste, a string
        of text.
<dbrock> Twey: don't misunderestimate series of squiggles
<Twey> dbrock: Yes, I was referring to the sumti itself, not the
       text representing it.
<dbrock> (to me, YOU are a series of squiggles)
<Twey> I am represented by a series of squiggles
<dbrock> Twey: the bridi itself, you mean?
<Twey> I am not the squiggles themselves
<dbrock> Twey: agreed (I was joking)
<xalbo> I am misunderrepresented by a series of squiggles.
<dbrock> haha
<Twey> Aye, but there was a serious point there too
<xalbo> The question is whether bridi1 is du'u1 or du'u2
<dbrock> xalbo: yes, I think you should speak up more
<Twey> dbrock: There were both a sumti and a bridi (or possibly
       two sets of squiggles representing them) in the question
       you asked
<Twey> At no time during this conversation was I talking about
       the squiggles.
<xalbo> I can't read the definition of {bridi} as anything but
        du'u2, but I could probably be convinced that du'u1
        would be "better" (.i ku'i zmadu fi lo ka xamgu ma)
<Twey> They are irrelevant to logic, which is one of the reasons
       nobody usually makes the distinction between ‘is a piece
       of text’ and ‘is represented by a piece of text’
<dbrock> Twey: {broda} in {lo broda} is definitely not a sumti,
         nor a set of squiggles representing one
<xalbo> .i xamgu fi ma
<xalbo> I had a logic teacher who drove the entire class *crazy*
        by insisting on that distinction.
<dbrock> the gimste says very explicitly that bridi1 is a string
         of text
<Twey> dbrock: ‘lo broda’ ‘is’ a sumti.  ‘broda’ ‘is’ a selbri,
       and thus a bridi.
<dbrock> Twey: can we please keep the distinction?
<dbrock> `lo broda' is a string of text, as is `broda'
<dbrock> please, let's not use the "shortcut" that a string "is"
         what it represents
<Twey> dbrock: It also says that that string of text is a
       predicate relationship, which is ridiculous interpreted
       in the exceedingly and unnecessarily literal fashion you
       are advocating, and therefore must mean that it
       represents a predicate relationship.
<dbrock> not in this discussion
<Twey> A series of squiggles cannot *be* a
       predicate relationship.
<dbrock> Twey: agreed
<Twey> A predicate relationship is an abstract thing of logic
       and/or grammar.
<dbrock> but "x1 (text)" is explicit
<Twey> Yes, it is
<xalbo> unfortunately, "is a predicate relationship" is
        also explicit.
<dbrock> how do you know they're not using your "shortcut"
<dbrock> you've said yourself several times that a string of
         text "is" a predication
<dbrock> why shouldn't the gimste?
<Twey> That is what I'm saying it's doing.
<Twey> It's the only way to reconcile the two facts
       presented above.
<dbrock> exactly, so the gimste means that bridi1 is a string of
         text
<dbrock> xalbo: very much less explicit
<Twey> Hm, interesting.
<dbrock> so we all agree that the gimste claims that bridi1 is a
         string of text
<xalbo> .ie
<dbrock> and we seem to agree that {broda} is probably a bridi
         even in {lo broda}, since {bridi} has no place for the
         grammatical context
<dbrock> .i lo'u broda le'u bridi lo ka broda kei zo'e
<dbrock> (in other words, {broda} is a bridi, period)
<xalbo> I can almost agree, although in that case, "even in {lo
        broda}" is meaningless; "even thought it is a substring
        of {lo broda}" I could accept.
<dbrock> agreed
<xalbo> I don't have a good theory of text, though, so I suspect
        I'm confused.
<dbrock> no, I think that is a very good formulation
<Twey> I agree with this line of reasoning, though it seems
       counter-intuitive to me.
<dbrock> me too, very much
<xalbo> I feel like it is meaningful to talk about "the word
        {me}" as a distinct thing from "the word 'me'", or "the
        first two letters of 'meat'".  This leads me to suspect
        that I don't have as good a formulation of all of this
        as I thought.
<dbrock> hmm, good point
<dbrock> we have that in Lojban too
<dbrock> {zo broda} feels different from {lo'u broda le'u}
<dbrock> even though it's "supposed" to be the exact same thing
<Twey> .ie-ru'e
<xalbo> and {lu broda li'u} feels even differenter.
<Twey> .ie
<dbrock> yeah
<dbrock> but I think this is a finer distinction
<dbrock> the distinction between du'u1 and du'u2 is a very much
         more important one
<dbrock> {zo broda} and {lu broda li'u} seem about as distinct
         as {ni broda} and {jei broda}, by the way, zo'o zo'o
         nai :-)
<xalbo> as if we knew what *those* meant :)
<dbrock> so anyway I'm going to go out on a limb here and
         suggest that we all pretty much agree that not only
         would it be better and more lojbanic for bridi1 to be
         the actual predication, but that the sort of abstract
         intent of the jbocevni was probably for {bridi} to be
         more about predications than squiggles
<dbrock> xalbo: as to why it would be better, for one, you
         wouldn't need {ka'e fatci} any more
<dbrock> for another, it suddenly becomes very natural to talk
         about "the bridi" in {lo broda}
<xalbo> I assume that a similar transformation would occur with
        {sumti}, but I don't know what would actually be able to
        fill the new place.
<xalbo> for that matter, what fills bridi3?
<xalbo> (well, it's a sequence, of a sequence of what?)
<dbrock> anything can fill sumti1 if sumti1 is just a predicate
         argument
<dbrock> probably of sumti1's (i.e., not text)
<dbrock> .i lu ko'a broda zo'u ko'a sumti .i je lo du'u ko'a
         broda cu bridi lo ka ce'u broda kei vu'i ko'a
<dbrock> .i lu ko'a broda cu jufra je nai bridi
<dbrock> .i bridi .ie jufra .i ku'i na bridi
<dbrock> now, to talk about grammatical elements, we need a
         completely separate set of selbri
<dbrock> maybe based on one with the place structure "x1 is a
         grammatical element of type x2 in expression x3"
<xalbo> Not sure.
<dbrock> let's say {gerna zei pagbu}: x1=p1 is a syntactic
         element of expression x2=p2=g3, playing grammatical
         role x3 according to grammar x4=g1
<dbrock> or {genpau} for short
<dbrock> then we can have {bridi genpau}, {selbri genpau},
         {sumti genpau}, etc.
<dbrock> or {bripau}, {selbripau}, {sumpau} for short
<dbrock> maybe {taurpau} instead of {selbrika'e}
<dbrock> and {jufra} doesn't need -pau
<dbrock> .i lo'u {.i lo broda cu brode} le'u jufra .i lo'u {lo
         broda cu brode} le'u bripau lo go'i .i lo'u {brode}
         le'u selbripau lo go'i .i lo'u {lo broda} le'u cu
         sumpau ra
<dbrock> .i lo'u {lo} le'u gadpau lo'u {lo broda} le'u .i lo'u
         {broda} le'u selgadpau lo'u {lo broda} le'u .i ji'a
         lo'u {broda} le'u taurpau lo'u {lo broda} le'u
<dbrock> .i .ia so'a selgadpau cu taurpau
<dbrock> .i xu drani
<xalbo> drani la'a
<xalbo> I'm still not sure, though.
<xalbo> (and *{gadpau} has a VU-pair; {gadypau} would be right)
<dbrock> (ah, right)


Comments and opinions welcome.

-- 
Daniel Brockman
daniel@brockman.se