--
You received this message because you are subscribed to the Google Groups "Lojban Beginners" group.
To post to this group, send email to lojban-b...@googlegroups.com.
To unsubscribe from this group, send email to lojban-beginne...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/lojban-beginners?hl=en.
{ka'e} is semantically close more to {cumki} than to {kakne}:
http://jbovlaste.lojban.org/dict/ka%27e
{lo plise ka'e farlu} means "It's possible that apples fall" (a
possibility whose realization doesn't hinge on volition) rather than
"Apples are capable of falling" (a capability whose realization hinges
on volition).
{ka'e} works by virtue of {cumki}'s place structure like {pu} does of {purci}'s,
http://jbovlaste.lojban.org/dict/pu
therefore the following analogy is the case:
da broda pu de = da broda fi'o se purci de = lo nu da broda cu purci de
da broda ka'e de = da broda fi'o se cumki de = lo nu da broda cu cumki de
{ka'e de} means cumki2, the condition under which cumki1 -- (lo nu) da
broda -- is possible. {ka'e la .luk.} implies that {la .luk.} is a
named condition. (I personally don't subscribe to the view that a
person sumti cannot fill a place defined in terms of abstraction,
event, condition, etc., so whether {la .luk.} refers to a person or
not wouldn't be a problem for me; but you can at will make a person an
entity separate from an event with {tu'a}: {ka'e tu'a la .luk.},
"under a condition involving Luke".)
{ca'a} is defined as {fi'o jai fatci}:
http://jbovlaste.lojban.org/dict/ca%27a
Because of the disarranging nature of {jai}, the sentential role of
any sumti after {ca'a} will be inherently vaguer than with {ka'e}. But
you still can use {ca'a} with a sumti like other such TAG couplings.
mu'o
How does volition fit in here? Apples are innately capable of falling
just like all other material objects, even in a universe with no
volition.
I wonder how your sense of {ka'e}'s meaning squares with the CLL's:
http://dag.github.com/cll/10/19/ — note in particular the claim there
that {ro datka ka'e flulimna} is true even though some ducks actually
can't swim, and that {la djan. ka'e viska} might be true even if John
has been blind from birth.
If this is {cumki}, mustn't the cumki2's be counterfactual conditions?
Is {cumki} meant to be used with counterfactual conditions?
> {ka'e} works by virtue of {cumki}'s place structure like {pu} does of {purci}'s,
Is this your guess? It was my impression that there was virtually no
agreement about CAhA's meaning as sumtcita. We certainly can't use
{pu} analogies in general.
By "capable" I was referring to the definition of {kakne}, which is
what I was comparing {cumki} with. Arguably, kakne1 is a conscious
agent (analogous to gasnu1) who is capable of bringing about kakne2,
and that's where I think the unique utility of {kakne} most resides,
i.e. {kakne} is more useful if kakne1 is more than {lo jai cumki}, an
entity potentially involved in a potential event. kakne1, whether or
not we call it "volitional", must be some sort of agent. {lo plise cu
kakne lo nu farlu} doesn't necessarily mean that the apple itself can
fall; it means that the apple can cause a falling of something. The
usual sense of "apples can fall" would be more accurately expressed as
{lo nu lo plise cu farlu cu cumki} or {lo plise ka'e farlu}. (That
{ka'e} morphologically derived from {kakne} but semantically rests on
{cumki} while these two selbri aren't exactly interchangebale, is
confusing, but that's how Lojban currently stands.)
> I wonder how your sense of {ka'e}'s meaning squares with the CLL's:
> http://dag.github.com/cll/10/19/ — note in particular the claim there
> that {ro datka ka'e flulimna} is true even though some ducks actually
> can't swim, and that {la djan. ka'e viska} might be true even if John
> has been blind from birth.
According to how this CAhA is actually defined, {ro datka ka'e
flulimna} means {lo nu ro datka cu flulimna cu cumki}, which I'm not
sure would be true if there already is an actuality where some ducks
can't swim. What's meant by this example is presumably that every
organism with the duck genome has the innate capability to cause the
event of itself swimming, {ro datka cu kakne lo nu vo'a flulimna}. {da
ka'e broda} can't exactly substitue for {da kakne lo nu broda}.
> If this is {cumki}, mustn't the cumki2's be counterfactual conditions?
> Is {cumki} meant to be used with counterfactual conditions?
I don't think so. {cumki} can describe a future possibility, something
that's neither counterfactual nor factual at a particular point in
time. {(da'i) lo nu mi ba'o fengu cu cumki lo nu mi pu na'e citka lo
plise} (I would have possibly been angry if I hadn't eaten the apple
earlier) is a counterfactual possibility, while {lo nu mi ba fengu cu
cumki lo nu mi ca na'e citka lo plise} (I might / would / will
possibly be angry if I didn't / don't eat the apple now.) is not
counterfactual. (Note that cumki1 in either case is still a
possibility even if the condition is met. cumki2 qualifies cumki1 as
more than an entirely impossible event (as true in at least one
possible world) and less than an actual event.)
mu'o
Ah, do you take the Lojban<->Lojban definition ({fi'o se cumki}) at
jbovlaste to be canonical?
Unfortunately I don't think this is a safe strategy in general,
whether or not xorxes is right about {ka'e}.
What? That's just plain backwards. Is there a single other language in which the canon definition is in a different language?
There is room for disagreement and discussion on how {ka'e} is defined
in Lojban on jbovlaste, but the canonical definition should be written
in Lojban anyway. English dictionaries for English speakers aren't
written in Chinese or Swahili; why should a canonical Lojban
dictionary for Lojban speakers be written in English, especially when
one of its design goals is cultural neutrality?
> Unfortunately I don't think this is a safe strategy in general,
> whether or not xorxes is right about {ka'e}.
What exactly is unsafe about defining a word in a logical language?
On 21 April 2011 09:28, MorphemeAddict <lyt...@gmail.com> wrote:
> The canonical definitions are and always have been the English definitions.
> To check whether a Lojban definition is right, it's checked against the
> English.
The English definitions suffer from many mistakes & issues:
http://www.lojban.org/tiki/BPFK+Section%3A+gismu+Issues
We have to check these things too. Against what? Cross-linguistic
grounds. That's what Lojban definitions should be checked against as
well.
mu'o
It's impossible to define every word in a language in the same language
without circularity. There must be some definitions which refer to things
outside the language. All words for species of organisms: "cinfo", for
instance, is defined as any organism of breed x2 which belongs to the species
Panthera leo, which is in turn defined by a type specimen.
Pierre
--
La sal en el mar es más que en la sangre.
Le sel dans la mer est plus que dans le sang.
--
While it may be true that Latin is today defined in such a manner, I would be very surprised if it was when it was not a dead language.
to pu benji di'u fo lo mi me la.android. fonxa toi
mu'o mi'e.aionys.
On Apr 21, 2011 3:32 AM, "Alex Rozenshteyn" <rpglo...@gmail.com> wrote:
Latin.
On Thu, Apr 21, 2011 at 5:10 AM, Jonathan Jones <eye...@gmail.com> wrote:
>
> What? That's just plain backwards. Is there a single other language in which the canon definitio...
>
>
> On Thu, Apr 21, 2011 at 2:28 AM, MorphemeAddict <lyt...@gmail.com> wrote:
>>
>> The canonica...
--> You received this message because you are subscribed to the Google Groups "Lojban Beginners" group...
--
Alex R
--
You received this message because you are subscribed to the Google Groups "Lojban Beginners" group.
...
Every tense/modal cmavo is supposed to be convertible to {fi'o
SELBRI}, effectively yielding a Lojban-Lojban definition. If we
couldn't so define {ka'e}, either Lojban lacks explanatory vocabulary
or there's no sufficiently logical ground for the very idea of {ka'e}
to be expressed as a bridi component.
On 21 April 2011 14:21, Pierre Abbat <ph...@phma.optus.nu> wrote:
> On Thursday 21 April 2011 01:14:38 Jonathan Jones wrote:
>> Ideally, /every/ Lojban word should be defined in Lojban. And whenever a
>> word /is/, that most certainly /is/ the canonical definition.
>
> It's impossible to define every word in a language in the same language
> without circularity. There must be some definitions which refer to things
> outside the language. All words for species of organisms: "cinfo", for
> instance, is defined as any organism of breed x2 which belongs to the species
> Panthera leo, which is in turn defined by a type specimen.
Yes, circularity is unavoidable. Here's the OED definition of "lion":
"a large powerful animal of the cat family, that hunts in groups and
lives in parts of Africa and southern Asia. Lions have yellowish-brown
fur and the male has a nane (= long thick hair round its neck)."
Every single word in it is circularly defined in other parts of the
same dictionary. But the text as a whole still describes, for speakers
of the language, what "lion" is.
On 21 April 2011 12:58, MorphemeAddict <lyt...@gmail.com> wrote:Every tense/modal cmavo is supposed to be convertible to {fi'o
> 1) When is any language fully defined? Not even Lojban is fully defined.
> Being fully defined is not a well defined concept. This was discussed here
> several months back.
> 2) Many (all?) languages lack vocabulary for concepts in other cultures.
> Lacking words doesn't mean that a language (Klingon, in particular) can't be
> used to talk about those concepts.
SELBRI}, effectively yielding a Lojban-Lojban definition. If we
couldn't so define {ka'e}, either Lojban lacks explanatory vocabulary
or there's no sufficiently logical ground for the very idea of {ka'e}
to be expressed as a bridi component.
> On Thursday 21 April 2011 01:14:38 Jonathan Jones wrote:
>> Ideally, /every/ Lojban word should be defined in Lojban. And whenever a
>> word /is/, that most certainly /is/ the canonical definition.
>> It's impossible to define every word in a language in the same languageYes, circularity is unavoidable.
> without circularity. There must be some definitions which refer to things
> outside the language. All words for species of organisms: "cinfo", for
> instance, is defined as any organism of breed x2 which belongs to the species
> Panthera leo, which is in turn defined by a type specimen.
Here's the OED definition of "lion":
"a large powerful animal of the cat family, that hunts in groups and
lives in parts of Africa and southern Asia. Lions have yellowish-brown
fur and the male has a nane (= long thick hair round its neck)."
Every single word in it is circularly defined in other parts of the
same dictionary. But the text as a whole still describes, for speakers
of the language, what "lion" is.
It's easy to fix this "problem" for any dictionary as follows:
(1) Take any word that appears in a definition but not as an entry and
classify it as "primitive".
(2) Take any word that appears in its own definition and classify it
as "primitive".
(3) Pick any remaining entry, classify it as "well-defined" and
classify all still unclassified words in its definition as
"primitive".
(4) Repeat step (3) until you run out of entries.
You end up with a pile of words classified as "primitive" and another
pile classified as "well-defined". You can try all possible picking
orders for step (3) if you wish to minimize the set of primitives for
the dictionary, since different picking orders will result in
different sets of primitives.
It would be interesting to see how many primitives result from this
process for a typical dictionary, and how the resulting list of
"primitives" compares with Wierzbicka's.
mu'o mi'e xorxes
--
You received this message because you are subscribed to the Google Groups "Lojban Beginners" group.
To post to this group, send email to lojban-b...@googlegroups.com.
To unsubscribe from this group, send email to lojban-beginne...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/lojban-beginners?hl=en.
Yes, such primes may exist. However, editors of the canonical Lojban
dictionary would still be on the hook to define even the valsi
corresponding to those primes. I wouldn't expect the entry for {barda}
to be empty because of its semantic primitiveness. And we wouldn't
need non-Lojban words to define {barda}, just like the OED doesn't
resort to Chinese to define "big".
mu'o
Well, here's hers: http://en.wikipedia.org/wiki/Natural_semantic_metalanguage
I don't know if these are the same for every language- I assume so, since that's the point, yes?- but most of them are cmavo in Lojban.
I also don't think they're very well chosen, as a lot of them are easy to define without circularity.
2011/4/23 Jorge Llambías <jjlla...@gmail.com>
On Sat, Apr 23, 2011 at 10:16 AM, MorphemeAddict <lyt...@gmail.com> wrote:
>
> Circularity is avoidable. That's the whole point of Anna Wierzbicka's
> Natural Semantic Metalanguage, which assumes a small number (~62) of 'words'
> that can't be defined in terms of simpler words.
>>> Almost all dictionaries are unapologetic about circularity.It's easy to fix this "problem" for any dictionary as follows:
(1) Take any word that appears in a definition but not as an entry and
classify it as "primitive".
(2) Take any word that appears in its own definition and classify it
as "primitive".
(3) Pick any remaining entry, classify it as "well-defined" and
classify all still unclassified words in its definition as
"primitive".
(4) Repeat step (3) until you run out of entries.
You end up with a pile of words classified as "primitive" and another
pile classified as "well-defined". You can try all possible picking
orders for step (3) if you wish to minimize the set of primitives for
the dictionary, since different picking orders will result in
different sets of primitives.
It would be interesting to see how many primitives result from this
process for a typical dictionary, and how the resulting list of
"primitives" compares with Wierzbicka's.
mu'o mi'e xorxes
The ones I always found a bit surprizing were the pairs GOOD - BAD,
BIG - SMALL, A LONG TIME - A SHORT TIME, NEAR - FAR, when one in each
pair could easily be defined in terms of the other: OPPOSITE-OF GOOD,
OPPOSITE-OF BIG, OPPOSITE-OF A-LONG-TIME, OPPOSITE-OF NEAR.
They don't have OPPOSITE-OF as a primitive, but presumably it must be
definable in terms of primitives (although it's not easy to see how).
And even if you needed to use BAD, SMALL, A SHORT TIME and FAR to
define OPPOSIT-OF, it would still be more economical to have
OPPOSITE-OF as the primitive instead of the other four.
Also, couldn't A-SHORT-TIME and A-LONG-TIME not be just SMALL
FOR-SOME-TIME and BIG FOR-SOME-TIME?
On Sat, Apr 23, 2011 at 2:04 PM, Jonathan Jones <eye...@gmail.com> wrote:
Well, here's hers: http://en.wikipedia.org/wiki/Natural_semantic_metalanguage
I don't know if these are the same for every language- I assume so, since that's the point, yes?- but most of them are cmavo in Lojban.
I also don't think they're very well chosen, as a lot of them are easy to define without circularity.Really? Which ones? How would you define them?
--
You received this message because you are subscribed to the Google Groups "Lojban Beginners" group.
To post to this group, send email to lojban-b...@googlegroups.com.
To unsubscribe from this group, send email to lojban-beginne...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/lojban-beginners?hl=en.
1) obvious misspellings:
CELLO "muscial*" (Not in OSPD3 -- RB)
CIVICISM "govermment*"
CLAMMING "particple*"
DUSTIEST "superative*"
HEADSAIL "fo*"
HEREDITY "chracteristics*"
MENACE "theaten*"
MURDER "premediated*" (I like it :-)
REDBIRD "plummage*"
SCREECHY "screaching*" (Not in OSPD2 -- RB)
THEIST "belives*" (Not in OSPD3 -- RB)
WORKBAG "instuments*"
2) words part of multiword phrases only ("multiword*" -- RB :-)
ABVOLT, KILOVOLT, MEGAVOLT, VOLT, VOLTAGE "electromotive#"
BREN "submachine*"
CANCHA, CESTA, FRONTON "jai*","alai*"
GILBERT "magnetomotive#"
KILOBASE, MUTON, REPLICON "nucleic*"
NONCOM, SERGEANT "noncommissioned*"
RIBOSOME "ribonucleic*"
SOUVLAKI "shish*"
VASOTOMY "deferens*"
3) words that basically should be in there, but are not...
AGEDLY "oldly*"
ART "esthetically*" ("aesthetically" is acceptable -- RB)
CITREOUS, CITRON "lemonlike*"
DRAGON "serpentlike#"
EFFLUVIA "byproducts*" ("by-products" in MW10 -- RB)
EYESTALK "stalklike*"
GENOISE "spongecake*" ("sponge cake" in MW10 -- RB)
HODAD "nonsurfer*"
HORN "bonelike*"
JACKDAW "crowlike*"
LANCE "spearlike*"
LATENTLY "dormantly*"
LEGEND "unverified#"
LEMMING "mouselike*"
MACE "clublike*"
MUMBLE, MURMUR, MUTTER "unclearly#"
QUAGGA "zebralike*"
SAC "pouchlike*"
SCOOP "spoonlike*"
VIROID "viruslike*"
Now back to your regularly schedulaed lojban.
--gejyspa
It's easier to keep I and YOU than forever distinguish between I as speaker and YOU as listener.
Besides, none of "speaker", "listener", "speak", "listener", or "-er" (meaning 'one who does X') is part of NSM.
And if the other person is currently speaking, then he is not I.
--
mu'o mi'e .aionys.
.i.e'ucai ko cmima lo pilno be denpa bu .i doi.luk. mi patfu do zo'o
(Come to the Dot Side! Luke, I am your father. :D )
--
mu'o mi'e .aionys.
.i.e'ucai ko cmima lo pilno be denpa bu .i doi.luk. mi patfu do zo'o
(Come to the Dot Side! Luke, I am your father. :D )
Oh, yeah, and I suspect that one of the main reasons that Lojban is not the canonical language of Lojban definitions is that their are no native Lojban speakers.stevo
Well, no. I refer you to CLL 17.14:
> Using any vocative except ``mi'e'' (explained below) implicitly defines the meaning of the
> pro-sumti ``do'', as the whole point of vocatives is to specify the listener, or at any rate
> the desired listener --- even if the desired listener isn't listening! We will use the terms
> ``speaker'' and ``listener'' for clarity, although in written Lojban the appropriate terms
> would be ``writer'' and ``reader''.
So do doesn't have to be even listening. And we also know that
mi'e can assign a plural to "mi". (CLL 7.2) Questions like what does
"mi'e le gerku" mean (presumably "okay, I am speaking for the dog
now") are left open.
--gejyspa
On Wed, Apr 27, 2011 at 2:28 PM, MorphemeAddict <lyt...@gmail.com> wrote:
It's easier to keep I and YOU than forever distinguish between I as speaker and YOU as listener.
Except that I /is/ /always/ the person speaking, and you /is/ /always/ the person listening.
"mi" is "lo cusku be dei" and "do" is "lo te cusku be dei".
"dei" is "lo nau se cusku", so I would pick "nau" as the primitive for deictics.
--
You received this message because you are subscribed to the Google Groups "Lojban Beginners" group.
To post to this group, send email to lojban-b...@googlegroups.com.
To unsubscribe from this group, send email to lojban-beginne...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/lojban-beginners?hl=en.
http://www.blackpool.gov.uk/EmailDisclaimer/
This message has been scanned for inappropriate or malicious content as part of the Council's e-mail and Internet policies.
Click here to report this email as spam.
http://www.blackpool.gov.uk/EmailDisclaimer/
This message has been scanned for inappropriate or malicious content as part of the Council's e-mail and Internet policies.
******************************************************************************
See the Blackpool You Tube video aimed at attracting French visitors by
clicking this link http://www.visitblackpool.com/jetaime
******************************************************************************
If I were to quote you as saying, "I love muffins.", the person that "I" refers to in that is you, because within the quote, you are the speaker. So yes, "I" and "you" do switch.
In the sentence, "You said it to me earlier.", the person being addressed is "You", and the speaker is "me", obviously.
If it were, "You said to me, 'I like muffins.', earlier.", both "You"and "I" refer to you, because within the context of the muffin quote, the speaker is you, while outside the audience is you.
to pu benji di'u fo lo mi me la.android. fonxa toi
mu'o mi'e.aionys.
On Apr 28, 2011 9:38 AM, "Michael Eaton" <michae...@blackpool.gov.uk> wrote:
So what happens if YOU address ME? Do YOU beome I, while I become YOU? Even if written from MY perspective? Consider the context of an autobiography: if somebody addresses the biographer, I is no longer the speaker (despite the fact that you are reading their presentation) as they are the person being addressed, and therefore the listener. At least from one standpoint, anyway.Also, you wouldn't want to introduce problems with sentences such as "You said it to me earlier". I and YOU surely cannot be always considered to be in a fixed role of speaker and listener.
>
> -----Original Message-----
> From: lojban-b...@googlegroups.com [mailto:lojban-beginners@g...
> --
> You received this message because you are subscribed to the Google Groups "Lojban Beginners"...
http://www.blackpool.gov.uk/EmailDisclaimer/
This message has been scanned for inappropriate or malicious content as part of the Council's e-mail and Internet policies.
Click here to report this email as spam.
http://www.blackpool.gov.uk/EmailDisclaimer/
This message has been scanned for inappropriate or malicious content as part of the Council's e-mail and Internet policies.
******************************************************************************
See the Blackpool You Tube video aimed at attracting French visitors by
clicking this link http://www.visitblackpool.com/jetaime******************************************************************************
--
You received this message because you are subscribed to the Google Groups "Lojban Beginners" gr...
That reference agrees with me, gejyspa.
On Thu, Apr 28, 2011 at 4:24 AM, Michael Turniansky <mturn...@gmail.com> wrote:
On Wed, Apr 27, 2011 at 4:45 PM, Jonathan Jones <eye...@gmail.com> wrote:Well, no. I refer you to CLL 17.14:
> On Wed, Apr 27, 2011 at 2:28 PM, MorphemeAddict <lyt...@gmail.com> wrote:
>>
>> It's easier to keep I and YOU than forever distinguish between I as
>> speaker and YOU as listener.
>
>
> Except that I /is/ /always/ the person speaking, and you /is/ /always/ the
> person listening.
>
> Using any vocative except ``mi'e'' (explained below) implicitly defines the meaning of the
> pro-sumti ``do'', as the whole point of vocatives is to specify the listener, or at any rate
> the desired listener --- even if the desired listener isn't listening! We will use the terms
> ``speaker'' and ``listener'' for clarity, although in written Lojban the appropriate terms
> would be ``writer'' and ``reader''.
So do doesn't have to be even listening. And we also know that
mi'e can assign a plural to "mi". (CLL 7.2) Questions like what does
"mi'e le gerku" mean (presumably "okay, I am speaking for the dog
now") are left open.
--gejyspa