From - Tue Nov 12 11:52:56 1996 Reply-To: Logical Language Group Date: Tue Nov 12 11:52:56 1996 Sender: Lojban list From: Logical Language Group Subject: Re: CONLANG: CONLANG IDEAS: for development X-To: conlang@diku.dk X-cc: lojban@cuvmb.cc.columbia.edu To: John Cowan X-UIDL: 87fb88db8e4220348c2cc37b17d8872f X-Mozilla-Status: 0011 Content-Length: 9355 Message-ID: <1rm1D4JzdoD.A.9EC._60kLB@chain.digitalkingdom.org> >From: blahedo@quincy.edu (Don Blaheta) >Subject: Re: CONLANG: CONLANG IDEAS: for development > >Quoth Bruce R. Gilson: >> Each full word, then, can be understood as noun, verb, or adjective, but the >> rules for relating these are easy to see from these examples: >> >> As noun As verb As adjective >> >> green thing to be green green >> house to be a house which is a house >> one who sings (singer) sing singing > >Hm. This sort of thing (trying to define "regular transformations" so >you can use one root as any part of speech) has a few problems. It has been proven workable with Loglan/Lojban (more so with the Lojban version, as there is very little TLI Loglan that makes use of the equivalence of parts of speech. >For instance, how would you express "greenness"? For the above transformations (which are really ENGLISH transformations) there is no change in the Lojban. ta crino That is a green thing. That green-s. That is greenish. ta zdani That is a house. That houses. That is hous-ish. ta sanga That is a singer. That sings. That is singer-ish. "Greenish" in Lojban is the "property of being green", one of several abstractions of the basic concept. ko'a ka ta crino #1 is (green-ness/the property of being green) #1 is the property of (That is a green thing). #1 is the property of (That green-s). #1 is the property of (That is greenish). > Or "song"? Two meanings. Lojban words are not all intransitive, so "song" is among other things the second place of "sanga" - one of the "objects" under English grammar terminology. ta sanga ko'e That is a singer of #2 (a song) That sings #2 That is (singing #2)-ish. But the "object" can come first under simple transformation with identical meaning to the above, or it can be paraphrased differently in English (but the above could be similarly paraphrased) ko'e se sanga ta That is a singer of #2 (a song) #2 is a song sung by that. That sings #2 #2 is sung by that. That is (singing #2)-ish. #2 is (sung-by-that)-ish. The act of singing is "song", and is a different kind of abstraction (or actually one of several kinds: ko'i nu ta sanga #3 is song/an event of (that) singing. (Omit the "ta" for generic "song".) and of course #3 is an event of (that being a singer). #3 is an event of (that singing). #3 is an event of (that being singer-ish). But there are 4 "subcategories of 'nu'": pu'u process (event) abstractor; x1 is process of [bridi] proceeding in stages x2 za'i state (event) abstractor; x1 is continuous state of [bridi] being true zu'o activity (event) abstractor; x1 is abstract activity of [bridi] composed of x2 mu'e achievement (event) abstractor; x1 is the event-as-a-point/achievement of [bridi] which modify the above with different flavors: ko'i pu'u ta sanga #3 is song/a process of (that) singing. (looks at a song as an evolving thing) ko'i za'i ta sanga #3 is song/a state of (that) singing. (looks at a song as a steady-state thing you start singing and continue till stopping) ko'i zu'o ta sanga #3 is song/an activity (that) singing. (looks at a song as broken down into numerous subevents - singing individual notes/bars/phrases, as it were) ko'i mu'e ta sanga #3 is song/a point-event of (that) singing. (looks at a song as a brief point event in the long view of mankind or of the life of singer "ta" or of the life of the song that is sung). >Or "sung"? One meaning was covered there - in the "se" conversion above. Another is based on the simple past tense: ta pu sanga That sang/was a singer/was singer-ish. ko'e pu se sanga ta #2 was a song of that (singer) #2 was sung by that. #2 was (sung by that)-ish. and we start exceeding the capability for English to follow the semantic manipulations. >Given a certain root, there are a number of ways to interpret it as a >different part of speech. Esperanto (at least as explained in the TY >series) absolutely infuriates me on this count--its words have gender! >Martel- (hammer) is fundamentally a noun, so we have "hammer", "to >hammer" and "hammer-like". One could say this about Lojban, but it works under tha same analysis as above. For the English verb "hammer" with an agent, you would make the compound "hammer-use". >But Sxovel- (shovel) is fundamentally a verb, so we have >"act-of-shoveling", "to shovel", and "shoveling". >From the Lojban perspective, "canpa" refers to what you are calling the "fundamentally a noun" version. The "fundamentally a verb" word corresponding most closely to Eo's is "cnapli" (shovel-use(r)), but it interprets a little differently than the above. ta cnapli That is a shoveler ... using an unspecified shovel. That shovels ... using an unspecified shovel. That is shoveler-ish (shoveling). which suggests that Eo's "noun" interpretation is inconsistent. "Act of shovelling" in Lojban would of course be the event abstraction or one of its subtypes as described above. >Eo requires suffixes in each case to get other meanings >"act-of-hammering" and "a shovel" (martelado, sxovelilo). Yet the only >way to know that sxovel- is a verb but martel- is a noun, is to memorize >it. Aagh. That argument could be used about any English word that has a predilection to be associated with a certain part of speech. It is the imperfect association with English meanings that you are finding irregular, because English "shovel" has two weakly-related meanings: the tool and the act of using such a tool. That is a flaw of English, not of Eo (though as I showed above, Eo is indeed inconsistent in a somewhat different way than you describe). If English "song" (the "tool" used by the singer") were the same word in English as the using of a song (i.e. "sing") you would say that Eo is arbitrarily choosing the verb form there. If English "green" also meant the "*method* used to produce the observation of the color green (such as emiision of green light or absorption of non-green light)", then you would see ambiguity there as well. But it is English ambiguity you are seeing, not that of the conlang. Other languages might distinguish between the shovel tool and the using of that shovel, and to them Eo would appear no more arbitrary about that word than any other. >I guess all I'm saying is, whoever runs with this concept, *be careful*. >It's all fine and good to have roots function as multiple parts of >speech (esp. for an IAL, since even if you screw up you'll probably be >understood), but it may not be as simple as you think. The lack of simplicity is primarily becasue the speakers are so used to thinking in terms of the parts of speech. Perhaps a language like Chinese that has less strong associations with the European parts of speech would seem equally arbitrary to you, and they would have less problem with Lojban's unitary "part of speech". What Don is calling the "gender of a word" - its predilection for nounal or verbal or adjectival usage - is really just an artifact of mapping the meaning of the word itself to English usage, and the 'genderness' is therefore really that of English. In the final analysis, ALL word meanings are essentially arbitrary. Conlang developers tend to try to devise meanings that are "useful". But the root meanings do indeed have to be individually memorized within the context of the semantic space of the conlang. Any language that does not cause the problem that Don notes with "shovel" is simply an encoded English. The legitimate complaint is that the conlang should consistently derive the affixal changes to the root meaning. In Eo's case that would mean that the -o ending would ALWAYS give the noun that forms the subject of the verb expressed by the -i ending (or some similar convention), and -ado and -ilo would have similar conventional modifications that would similarly always have the same correspondences to other affixes. This type of generalization works in Lojban (though Lojban has no suffixes but relies on context to give any part-of-speech flavor), and presumably could work in Bruce's concept. Defining the effect of various affixes invariably in terms of each other solves the "problem" of inherent 'genders' of roots. (I think this agrees with Edmund for the most part, but I read his post after writing this.) ---- lojbab lojbab@access.digex.net Bob LeChevalier, President, The Logical Language Group, Inc. 2904 Beau Lane, Fairfax VA 22031-1303 USA 703-385-0273 Artificial language Loglan/Lojban: ftp.access.digex.net /pub/access/lojbab or see Lojban WWW Server: href="http://xiron.pc.helsinki.fi/lojban/"