very intemperate response to Lojbab on Fifafuban

[ucleaar <ucleaar@UCL.AC.UK>]

Summary: Lojbab says fifafuban is too hard to ever be acceptable
Lojban, because it is too complex and too unfamiliar. And replies
that it is demonstrably not too complex and while it is unfamiliar
by today's standards, basing prescriptions on today's standards
is like basing standards for adult conduct on norms of children's
conduct.

> >> And you rearranged the terms for no apparent reason except to be
> >> obfuscatory.
> >I did no *rearranging*.  I did only *arranging*.  There are no rules
> >concerning reasons for possible arrangements.
> There is the implicit rule that any deviation from the shortest, most
> unmarked form of expression should have SOME reason for being used.  I
> think that this is a Gricean maxim of some kind, as well as perhaps a
> corollary of Zipf's Law.

Truish, but not yet true of Lojban. You don't judge norms of French style
by the half-baked attempts of foreign learners to use it. And more
generally, the Gricean approach is good on communication but weak on
style; not all aspects of style are attributable to straightforward
communicative issues.

> Among the recognized reasons for non-standard sumti order are 1) emphasis
> on one particular sumti by moving it (alone) out of its normal position;
> 2) relocating a sumti to avoid an inconveniently "heavy" sumti before
> another sumti, or perhaps before the selbri (this is the explanation I
> favor for "cumki fa lenu ..."); 3) fronting a sumti for parallelism with
> a description sumti - this usually involves using SE conversion as well;
> 4) matching a translation or native language word order structurally
> literally; 5) poetry - i.e. to match a certain rhyme or rhythm or
> word/sound-pattern scheme.

It's a good list, but unless it's prescribed by the refgrammar I should
and will try to ignore it. Inevitably initial usage, starting from no
established patterns, and with glico intuitions, will vastly underuse
the wide range of resources the grammar already affords. Notions derived
from such jejune usages, such as what is "non-standard sumti order" and
which reasons for it are "recognized", should be scrupulously paid no
attention to.

> The movement out of numerical order of more than one sumti is probably
> not justifiable except for reasons 4 or 5. This doesn't mean it is
> illegal, but it IS a hindrance to communciation.

Fluent speakers of a language, when addressing nonfluent speakers, tend
to turn their own speech into pap, that the addressee may be the better
able to comprehend it. Simplification to the point of virtual breakdown
may assist communication in such contexts, but should hardly be taken
as a model of good usage.

> >Current Lojban usage has a style like you'd expect of a children's book
> >- as simple as possible.  Naturally most present users value easy
> >communication more highly than anything else, but I was attempting to
> >dip my toe into the expressive powers of less well visited areas of
> >grammar.
> Nora's textbook writings and examples are children's bookish.  The stuff
> written as part of the Chinese whispers is almost certainly not.

I don't know how to assess that objectively. My impression, having read
and participated in them, is that it is childish in its syntactic
complexity, and worse than that in its general clarity. Nick is of the
view that such simple syntax is a great virtue; he thinks that in
lojban, levels of subordination found in even an everyday english
sentence are intolerable. He may be right. But it's worth experimenting.

> Rearranging sumti seemingly at random has NO apparent expressive
> purpose, and that is precisely why it is stylistically dead.

They're not rearranged. They're merely arranged. The sumti must be in
some order. That they are usually given in 1-2-3-4-5 order is mere
habit, mere laziness, or mere caution.

> >Your questions imply that fi-fa-fu is analogous to OSV in English and to
> >passive.  It isn't.
> I agree - the English usages have stylistic meaning. fi-fa-fu does not.
> There is neither recognized convention nor implicit Gricean or other
> linguistic basis to give reader any idea what you intended by the
> placement of each sumti.

Right. And this absence of convention means the writer can proceed with
impunity. I did only what was permitted by the grammar, and nothing I
did clashed with any rules of information structure.

> In the meantime, you stretched the mental juggling of human parsing by
> forcing people to keep too many slots unresolved - your example was a
> sort of center-embedding of a 5-place predicate in the middle of an
> unresolved main bridi, where in addition you rearranged the terms of
> that center-embedded subordinate description.

This is rubbish. As can be demonstrated by constructing a syntactic
analogue in, e.g., english, this does not stretch the mental juggling
of human parsing. It merely stretches the feeble competence in Lojban
that you and everyone else so far has.

> Now the fact that in addition you used the still-unusual constructions
> of "jai" and "me" added insult to injury, but I do consider their usage
> to be a valid sort of stylistic experimentation.

oioioioioi! "insult to injury"! Why take the trouble of designing them
into the language, if you don't want them used? Why does using certain
cmavo amount to "stylistic experimentation"?

> Looking at your text again:
> > >   [fi la pou lojbab ralju]
> > >   [fe lei jai fau skicu
> > >       [be fo lo jbovla]
> > >       [bei fe maa]
> > >       [bei fai ro da poi kea me maa]
> > >       [bei fi da]]
> > >   [fa diu]
> > >tinbe
> The result of your usage included a string of 16 consecutive cmavo.  I
> dare say it would not be easy to write an understandable English
> sentence that used 16 consecutive English function words (i.e. words
> that are not nouns/verbs/adjectives/adverbs).  Lojban WILL have more
> cmavo than English, but I think it still requires SOME content to be
> understandable.

Could you, but not each of us, ever, with so very much of this and so
very little of that, so much down here or so little up there, ....

- embed that in some suitable context and there's not an iota of difficulty.

You're making up these observations as you go along. What can I say?
In this string of 16 cmavo there most certainly is "content". Long
sequences of cmavo are difficult? Show me the evidence.

> >It is usually not obfuscatory to say something in more words than
> >necessary.  It is commonly asserted that is is obfuscatory to say
> >something in more words than necessary, but the commonness of this
> >assertion is attributable to widespread ignorance about language rather
> >than to its essential truth.
> >In (1) the presence of "some" does not increase difficulty, while in
> >(2-3) the presence of "that" reduces difficulty.
> >(1) I found (some) coins down the back of the seat.
> >(2) I doubt (that) it will.
> >(3) I met the man (that) she had been telling us about.
> Adding one word is quite understandable and acceptable.  Adding words
> and changing order is less so
>   "Down inside the back which was a part of the seat some coins were by
>   me found."
> would probably be considered stylistically unacceptable English.

I see no problem with that. The preposing of "by me" probably adds extra
syntactic complexity, and for this or other reasons is marked, but it's
still okay.

> Likewise
> "That it will is doubtful to me".

There exist conventions in english that disfavour clausal subjects.
There is nothing analogous in Lojban, - if, that is, we go by the
refgrammar documentation rather than by whatever rules you care to
pluck out of the air to justify your prejudices.

> or
> "There was this man that she had about him to us been telling that I met."

Do you think that "She had been telling about him to us that I met this man"
is normal? I don't understand your example.

> Your usages seem to me as strainedly abnormal as these.

I reject the basis on which you form your impressions. In English there
exist conventions of style and usage. Not in Lojban. Further, in most
cases in english the syntactically more marked construction is also
syntactically more complex. This is not so in Lojban.

> >> I speak English quite well. but I guarantee that there are a lot of
> >> usages for which I cannot expressly state the rule I use to justify
> >> a given usage
> >I venture that there is no usage for which you can expressly state the
> >rule.  That does not matter.  What matters is that you know the rule,
> >and we can tell that you know the rule because we can observe you using
> >it.  Working out what the rule actually says is what keeps academic
> >linguists employed.
> A ball falls to the ground.  By the above logic, the ball "knows" the
> law of gravity.

I don't see that. The ball falls because acted on by gravity. There is
no external force acting on your usage; or at least that is a reasonable
assumption.

> The rules we attribute to English are a model of the way our minds
> behave when faced with examples that we label with the name of the
> language.  It is NOT clear that the model is reality,

It is not clear that any model or anything is reality. This, therefore,
is a pointless objection.

> that our minds recognize anything remotely like the rules in the model,
> or even that the rules in our minds have any linguistic properties at
> all per se, as opposed to being representations general pattern-matching
> rules that govern biochemical interactions, that in turn cause what we
> call thinking and/or speech.  The rules may be (and I think ARE) an
> illusion, and probably at best only an approximation.

If a machine is able to say whether any given sentence is or isn't
grammatical, then it shall count as knowing the rules of grammar. How
that knowledge is represented is another question. Thus I don't accept
that there's even a prima facie case for thinking the rules are illusory.
[Mind you, I personally feel that it doesn't matter that much to
linguistics whether they're illusory. But this is also another matter.]

> In the case of a conlang, the prescription is indeed a governing set of
> rules. But we already know that actually Lojban use does not always
> follow the prescription. I don't claim that you are not writing Lojban
> when you leave out the apostrophes, do I?.  Nor, even if I misunderstand
> some experimental cmavo that you have proposed that I have not learned,
> can I justifiably say that you are not speaking Lojban in using that
> experimental cmavo.

When we judge which language a text is in, we are asking which set of
linguistic rules is intended by the speaker to be used in decoding the
text. We don't examine the text and ask "for which language is this text
grammatical" - often there is no known language for which the text is
grammatical. The upshot is that actual lojban use has no bearing on the
issue of what the lojban rules are.

> >> Well, there OUGHT to be a brivla for every member of ZAhO.
> >> And tanru are supposed to be - tanru, and NOT "rule-governed" in
> >> semantics.
> >That's why I object to their being suggested as a solution.
> Since Lojban semantics as a whole are explicitly NOT "rule-governed",
> then you object to all of Lojban as a solution.  %^)

Lojban semantics is mostly not rule governed because it is mostly
nonexistent. That part of lojban semantics that exists is rule
governed.

> >> There is little justification for loading up selbri with all manner
> >> of reducers of semantic ambiguity
> >The justification for loading up selbri with "all manner of reducers of
> >semantic ambiguity" is both great and obvious: it reduces vagueness.
> But in Lojban, degree of vagueness is optional.  In English, people have
> to STRETCH the language, and even to misuse it (by having words mean
> other than their denotations) in order to be vague (listen to any
> politician for an example).  In Lojban, the vagueness is built in, and
> you have to be more wordy in order to be more precise.

Give examples. None of what you say here seems true to me.

> >> Because people memorize the places in a certain order rather than as
> >> having certain place numbers, and you are intentionally marking them by
> >> their place numbers.  This forces the listener to tag the sumti by place
> >> number and overtly rearrange them in his head for interpretation per the
> >> selbri place structure.
> >If true, then this would indeed be a reason why fi-fa-fu is hard.
> >But if true, it means Lojban is syntactically unlike natlangs, which
> >indeed it may be.

I should add here that my claim that distinguishing "arguments" by order
is unlike natural languages is theoretically controversial. In some theories
complements are distinguished by their position on a list, though i'm
not sure if they're distinguished by that alone.

> A cigar for the gentleman!  Since no natlang has a predicate grammar, of
> course Lojban is syntactically unlike natlangs.

What is a predicate grammar? In which respects is Lojban syntactically
unlike natlangs?

> But this is irrelevant. Languages with strong word order rules may
> technically tolerate violations of that word order and still be
> understandable, but it is also possible that they will NOT be easily
> understood even by a native speaker.  A poetic sentence like "To the
> store via the main drag go I". is understandable, but at some level of
> complication a similar sentence structure would cease to be
> understandable even to a native speaker in real time.

This is an underresearched area, so obviously you're inventing this as
you go along. But nonetheless it does seem reasonable to suppose that at
some level of complication a structure ceases to be processable even to
a native speaker. It should be possible to measure this. But anyway,
no sentences we've been discussing would stretch the processor to breaking
point.

> fi-fa-fu Lojban is theoretically analyzable on paper, just as 10-deep
> center embeddings in English.  But whether either will work for the
> language in real usage is questionable enough that the rules perhaps
> should NOT exist in an infinitely recursive form.  In natlangs, I
> believe that the "rules" do NOT work with infinite recursion.

Is it a rational belief? What is an example of a rule that won't work
with infinite recursion?

> >None of this stuff you're on about is in the refgrammar.
> Thereby suggesting that it is not necessarily part of the standard
> language (or that John needs to add to the material %^)

Exactly. This is one of the points I've been making.

> >My purpose is not to be obfuscatory or ornery, or at least not for the
> >hell of it.  Lojban is defined as having syntactic capabilities far more
> >complex than those that actually see use.  What is actually used is
> >pidgin lojban.
> By that logic, then what gets used in real life is "pidgin English"
> because deeply center-embedded sentences are not used or understood.

If we could agree on some measure of syntactic complexity appropriate
to both english and lojban we would find that the lojban sentence was
by english standards not difficult. I have already illustrated this.
The lojban sentence is not deeply centre embedded - at least not by
english standards.

How is it that you can read lojban text only with considerable effort
(as evidenced by your failure to read it when posted on lojban list)
yet you assume that when you find a lojban sentence too difficult it
is not because you do not have good command of the language but
rather because that sentence would cause breakdown in even the most
competent reader?

> >If you find the result difficult, it shows either that you just don't
> >have a good enough command of Lojban, or that there are real parts of
> >lojban grammar that are determined but not yet promulgated in documents,
> >or that the design doesn't work.  Personally, I think it is most likely
> >that the first of these is so, and least likely that the last is so.
> Whereas I think that there are pragmatics and semantics rules that will
> apply to Lojban as used, rules that will deviate from the prescription
> and which will NOT be documented (and perhaps not "determined").

This is just equivocation about the meaning of "Lojban". If we mean
Lojban-as-defined-lojbab then I fully accept that my fifafu sentence
is execrably unprocessable lojban.

> The level of nesting of Lojban's theoretically infinite nested
> structures is an example of such a rule that will never be prescribed;

Do you consider this a semantics rule or a pragmatics one? Why should it
be either?

> so are limits on rearranged orderings of sumti.  But the worst that one
> can claim out of this is a parody of Chomskyism - that no Lojbanist will
> ever ever have fluent "performance" that matches their "competence" in
> the language.

This claim would make no sense. What would performance that "matches"
competence be? [where I assume "competence" is meant by you in Chomsky's
sense]

> Are no English speakers fluent because no English speaker can understand
> a 10-level center-embedding?

No, I'd have thought. Your question implies someone would think Yes. Who?

coo; mie and