[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: expansion to sentences
Bob Chassell writes about expanding a <BAI> first to a restrictive
subordinate clause and then to a separate sentence. I am happy to see
this kind of exploration of semantic transformation.
It is possible to look at the same thing from different aspects, and
I don't want to say that Bob is "wrong" because he isn't. But in many
of the things I do with brivla, I find it convenient to stick with the
Old Loglan view (per JCB, undoubtedly derived from some classical
philosopher) that a brivla is a relation between not only the occupants
of its numbered places but zillions of other possible modal places.
Where this comes in particularly handy is in descriptive sumti. Its
selbri is defined as a list of N-tuples of thus-related objects (where
N includes both numbered places and modals). To compute the referent
set, retain only definition list members that have the right
sub-arguments (e.g. lo se citka be le ratcu, only things eaten by the
specified rat). Then pull out the x1 occupants and those become the
full referent set. (It is then affected by le - lo - loi - etc.)
The idea of transforming <BAI> phrases to restrictive subordinate clauses
is very congenial to me. In -gua!spi in fact there are no modal
operators; sub clauses are short and efficient, and are used for this
service.
Bob asks whether a <BAI> exists which can effectively substitute for
<FA> on every place of every predicate. I agree that there probably
is one. When large sets of arguments share a common <BAI>, they
constitute a "case". Cases are common in Lojban, though we deny that
every place has to fit into a big case, and the numbering varies to
suit the particular word, not keyed by case. For this reason Lojban
infuriates case grammarians and also non-case grammarians, and everyone
keeps telling me to stop using the term "case" in connection with
Lojban.
-- jimc
> Date: Wed, 12 Jun 91 12:43:19 EDT
> To: ai-lab!lojban-list@snark.thyrsus.com
> From: grackle!bob@gnu.ai.mit.edu
> Subject: expansion to sentences
>
>
>
> Recently, I suggested that:
>
> Each of the members of BAI can be expanded into a form with fi'o
> and the member's associated gismu; each of these expansions can,
> in turn, be expanded into a relative clause with poi:
>
> ka'a zo'e -> fi'o klama zo'e -> poi zo'e klama
>
> Just now, I discovered the next step, which is to expand a sentence
> with a relative clause into two separate sentences.
>
> Here is an example:
>
> le zarci ka'a mi cu barda 1.
> le zarci fi'o klama mi cu barda 2.
> the market gone to by me is big.
>
> le zarci poi mi klama [ke'a] cu barda 3.
> the market which I go to [it] is big.
>
> le zarci goi ko'o cu barda ije mi klama ko'o 4.
> The market, X4, is big and I go to it.
>
> This exercise produces several insights.
>
> As la lojbab says
>
> BAI was formed with the intention that every place structure place
> could be labelled more or less accurately with one of [its
> members]
>
> But now that I have expanded modal phrases into sentences, I no longer
> think of BAI as creating new places or as labelling old ones. Instead
> I think of BAI as creating new predications about the referent of the
> first place of the brivla, or as uncovering aspects of the predicate
> that were hidden.
>
> Here is an example in which, conventionally, a member of BAI adds a
> place:
>
> mi cusku zu'i bau la lojban. 5.
> I express something-typical in-language Lojban.
>
> "bau la lojban." adds an "in-language" place to "cusku" which has the
> definition: "express/say...to...in form/media...".
>
> The sentence expands to:
>
> mi poi la lojban. bangu cusku zu'i 6.
> I who Lojban is-the-language-of express something-typical.
>
> This expands to two sentences:
>
> mi cusku zu'i 7.
> .ije la lojban bangu mi
>
> The new place becomes a predication about the first place of the
> original brivla!
>
> What is happening here? Am I confusing myself utterly, and
> misunderstanding Lojban, or have I found something interesting?
>
> I am beginning to think the latter: that this exercise tells us that
> it is useful to consider an `added' place as being a second
> predication in the sentence, in addition to the predication provided
> by the main selbri. The reason the `added place' is a second
> predication is that the first predicate did not encompass the notion.
> Indeed, if it had, the new place would not have been required.
>
> On the other hand, it is evident that "in language..." could well have
> been the fifth place of "cusku". In a polylingual society, the
> language in which something is expressed may be important. Earlier
> this century, for example, as the result of a political conflict among
> rural people, well educated people, and city people, Norwegian civil
> servants were required to answer letters in the same language as they
> were written.
>
> This discussion of new places helps explain the process of creating
> definitions: to add a new place to a definition is to merge a new
> predication into the existing one. This suggests that the meaning of
> a brivla can be resolved into more primitive parts. Moreover, when a
> member of BAI is used to label an existing place structure, the
> process serves to illuminate a predication that is already part of the
> definition of the brivla. In this case, BAI serves as a dissecting
> tool!
>
>
> Here is another example using BAI. In this example, the member of BAI
> is placed in a position that looks adverbial:
>
> mi bai limna lo korbi be lo lalxu le porpi bloti 8.
> I compelled-ly swim to the edge of the lake from the broken boat
>
> But I understand this as expanding to:
>
> mi bai zo'e limna ... 9.
> I compelled by something unspecified swim ...
>
> which I view as adding a new place to swim, the "compelled-by" place.
>
> The previous sentence can be expanded to
>
> mi poi zo'e bapli limna ... 10.
> I who something compells swim ...
>
> and to the two sentences
>
> mi limna 11.
> ije zo'e bapli mi
>
> I find it interesting that the expansions always seem to attach to the
> first place; the first place does appear to have a special role.
>
>
> Here is another issue: la lojbab says on page 6-17 of the draft
> lessons:
>
> ...a restrictive relative clause serves only to identify the
> relativized phrase, [therefore] it doesn't really affect the truth
> value of the main predication of the sentence, so long as it
> adequately restricts the relativized phrase.
>
> However, I interpret an utterance such as
>
> le zarci goi ko'o cu barda ije mi klama ko'o 12.
> The market, X4, is big and I go to it.
>
> as requiring both sentences to be true for the utterance as a whole to
> be true.
>
> But if this be the case, then "ka'a mi" in the following must be true
> for the original un-expanded sentence to be true.
>
> le zarci ka'a mi cu barda 13.
> the market gone to by me is big.
>
> This suggests that modal phrases are veridical, similarly to sumti
> with "lo".
>
>
> Finally, I have a question: do members of BAI exist to label all
> second and subsequent places of all the gismu? I think `yes'.
> Testing this hypothesis would be an interesting exercise for someone;
> and in the process that person would half write a lojban-lojban
> dictionary!
>
> Consider "lisri" for example:
>
> lisri sri story story about...told by...to...
>
> about... secu'u expressing (saying)
> or perhaps fau in the event of (non-causal)
> or even sera'a concerning (pertaining to)
>
> told by... cu'u as said by source (attribution)
> to... teja'o witnessed by (audience)
>
>
> That is all for now. I am looking forward to your comments.
>
>
> Robert J. Chassell bob@gnu.ai.mit.edu
> Rattlesnake Mountain Road (413) 298-4725 or (617) 253-8568 or
> Stockbridge, MA 01262-0693 USA (617) 876-3296 (for messages)