archives of the CONLANG mailing list ------------------------------------ >From robin@extro.ucc.su.OZ.AU Wed Feb 2 12:16:50 1994 Date: Wed, 2 Feb 1994 01:16:50 +1100 From: Robin F Gaskell Message-Id: <199402011416.AA14928@extra.ucc.su.OZ.AU> To: conlang@diku.dk Subject: Syntax Analysis #From: robin@extro.ucc.su.oz.au (Robin Gaskell) 5 Dec 93 #To: conlang@diku.dk (Conlang Mail List) #Subject: Syntax Analysis Hello Friends, Just a little something I whipped up. In his last letter to me, Haitao said that we would have to decide on the system of grammar we wanted for Glosa. So, I thought about it. Naturally, I had been waiting for the authors to decide that the time was right for them to deliver a tabulated version of Glosa's grammar. But it hasn't happened, and Haitao is a programmer, who could probably produce a parser for the language. I decided that this must be the time for a tabulated Glosa grammar to appear: if the authors weren't going to write it, perhaps I would have to do so. While casually looking at a book on meaning representation, I noticed that it was full of pseudo-algebraic formulations, and that most of the symbols would not transmit through the network as e-mail. Pretty silly, I thought; it would seem a lot better if our codification of syntax was able to network readily ... using ASCII symbols, obviously. So, GAS was born - gASCII Analysed Syntax: in my ignorance, I imagined that each syntactic function could be awarded a symbol, preferably a non-alphanumeric one, and, I thought, doing such an analysis should tell me something. To start with, I wondered how many syntactic functions there were: it was something like the earlier question about the number of types of verbs and nouns. We would need a rational number: not so many it was unmorkable; and not so few it was meaningless. Then I started to count the symbols on my 101 key keyboard: there were the normal 26 letters, each with upper and lower case versions; then there were the 10 numbers; and after that, I counted 31 non-alphanumeric symbols. I decided it would be nice if the syntactic categories could be restricted to this last group of 31 symbols, so they wouldn't get mixed up with the normal text. But, already, I've got up to 38 symbols plus another 7, which are two-symbol compounds: this yields a total of 45 syntactic categories. Maybe this is too many, and I'll have to slim it down a bit. Oh yes, I mentioned Haitao. Some months ago, I posted his `pidgeon post' address, in deepest China, on Conlang - reporting that he was eager to make contact with researchers in the West, who were looking into Natural Language Processing as it articulates with the IAL concept. Did anyone write to him? Do any of you know of anyone, at all, who might be doing NLP / IAL research? In a later blip, I will suggest that of all the countries in the world, China has most to gain from adoption of the IAL. So, here it is:_ The gASCII Analysed Syntax (GAS) system (Alpha Test ver.) ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Brief Outline. * This is a means of codifying the syntactic function of the elements of a sentence. Using this analysis, we can see the functional structure of a sentence, without needing to refer to the actual words. * The analysis can be done either manually or by machine; if done automatically, the code can be used as part of a program that parses the language. It has the advantage that it can be scrutinised, if necessary, during the parsing process. Natural Language Processing - This system was invented with both English and the Planned Language, Glosa, in mind. Although linguists have attempted to codify the syntax of English, the average educated person has learnt his or her use of syntax through practice, and there is no easily-read reference book in which the syntax rules can be found. - Glosa is in a worse position: prior to this, no-one has codified its intuitively-used syntax. This system is presented as a means of stabalising that syntax. Rationale. i. ASCII code can be read by OCR software. ii. ASCII code transmits through the Internet. iii. The various syntactic elements can be covered, in general, by the non-alphanumeric ASCII symbols. iv. Some semantic categories can also be shown. v. Small syntactic distances are shown using single spaces. vi. Syntactic spaces between phrases are shown as double spaces. vii. Larger syntactic distances, such as those between the major parts of a sentence (S-V-O) and between clauses, are indicated with triple spaces. viii. Clauses are marked with brackets - differently for adjectival, adverbial and noun clauses. ix. Non-literal language is marked: for ease of recognition and as an aid to machine translation. x. Patterns of syntax can be found by analysis, and used to prompt improvements in clarity. xi. Preferred patterns of syntax can be readily recognised and taught. xii. As part of a meaning representation system, the code (bearing the linguistic function) would be matched with the word or symbol (carrying the semantic content). xiii. Languages that have no morphological grammar will use this code to hold the information usually found in Part-of-Speech markers and grammatical inflections .. for purposes of the machine handling of information and translation. xiv. Generation of this code will be a function of Artificial Intelligence; the code generated can be perused and understood by the human operator of a mechanised translation system - thus allowing the process to be monitored. xv. The code permits the use of unchanging concept-words, ordered according to a syntax-based grammar ... in metalanguages, Intermediate Languages and concept-based auxiliary languages. gASCII Elements ~~~~~~~~~~~~~~ Basic . ! > @ $ substantive action modifier space, time logical (noun) (verb) (adjective, preposition preposition adverb) Tense / \ ~ ^ | future past continuous conditional now Modifiers # % > v number quantity quality auxiliary (countable) (measurable) (property) verb Conjunctions + & joins words, phrases structural: joins clauses Functions x t = < <` location time equals, like verb is participle X proper noun as , similar passive 0 ? - , ; negative: un- general joining pronoun pronoun no, not, never question concepts personal impersonal nothing (compounds) People o s ' `. other self possessive gerund O proper noun S name of 1st person Specific ?o ?. ?! ?x ?t questions who what why where when Clauses ( ) { } [ ] " " adjectival noun adverbial parenthesis or quotation Non-literal : * * _ _ language metaphor or idiom start end other n-l term of sentence >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<< Examples of GAS in application The cat sat on the mat. = U felis pa sed epi u tape. _. \! @._ Plu studenti fu memo: na pa dice de u Tesaurus de Roget plura kron. _o /! [, \! @.@O #t]_ Three fat boys sat by the river bank, and ate jam sandwiches. _#>o \! @.-. & \! >._ While three fat boys sat by the river bank, and ate jam sandwiches, their sisters stole their bicycles. _[t #>o \! @.-. & \! >.] ,'o \! ,'._ Tem tri paki ju-an pa sed proxi u ripa, e pa fago plu konfekti pani, mu plu fe-sibi pa klepto mu plu bi-rota. <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> Cheers, Robin ______________________________________________________________________ >From shoulson@ctr.columbia.edu Tue Feb 1 06:10:07 1994 From: shoulson@ctr.columbia.edu (Mark E. Shoulson) Received: from localhost (shoulson@localhost) by startide.ctr.columbia.edu (8.6.5/8.6.4.788743) id LAA23057; Tue, 1 Feb 1994 11:10:07 -0500 Date: Tue, 1 Feb 1994 11:10:07 -0500 Message-Id: <199402011610.LAA23057@startide.ctr.columbia.edu> To: conlang@diku.dk Subject: Planned languages server >Date: Tue, 1 Feb 1994 08:19:34 +0100 >From: helz@ecn.purdue.edu (Randall A Helzerman) >Could someone tell me what is the ftp address of the planned languages server? The Planned Languages Server is currently down indefinitely. When it was up, it never had an ftp site. There are conlang-related things available via ftp from world.std.com in /pub/lingua (not Lojban or Esperanto, chiefly), surplus.demos.su in /esperanto, rand.org (also Esperanto), casper.cs.yale.edu in /pub/lojban, and HablI.tamu.edu (Klingon, just starting up), all according to recent messages on this list and the Klingon list. I haven't tried most of these places, so I can't tell you how much is there. ~mark ______________________________________________________________________ >From shoulson@ctr.columbia.edu Tue Feb 1 06:47:25 1994 From: shoulson@ctr.columbia.edu (Mark E. Shoulson) Received: from localhost (shoulson@localhost) by startide.ctr.columbia.edu (8.6.5/8.6.4.788743) id LAA23316; Tue, 1 Feb 1994 11:47:25 -0500 Date: Tue, 1 Feb 1994 11:47:25 -0500 Message-Id: <199402011647.LAA23316@startide.ctr.columbia.edu> To: conlang@diku.dk Subject: Syntax Analysis Hmmm. Interesting method you have there, Robin. Is it really any better than actually using the words instead of symbols? e.g. instead of _. /! @._, maybe ? OK, it certainly takes much more typing, but it's easier to read. It's also language-dependent (whatever language you use for the words), and could be less useful for marking sentences for analysis (as in _the .cat /!sat @on the .mat_), though really not too badly: any computer worth its salt could be perfectly capable of pulling out the information in the <>'s and treating it separately. Though similarly, it could use the symbolic method and translate to the format (in any of several conventions/languages) for debugging output by the human operator. You should realize, of course, that your method is quite limited to languages that resemble English and Glosa. Not all languages have participles, linking verbs, etc., and not all languages can have their tense system so simply expressed as past/present/continuous/conditional/now. Some require perfectives, etc. Words like "gerund" and "passive" don't have any meaning in many languages, but instead other constructs which you don't treat do. This may not matter, but it will limit which languages can use this system. Also, I note that you don't mark direct objects. This will limit your method's usefulness to coding only those languages which mark case by word-order, and that in the same order as English/Glosa. Coding "a fire I saw" would not help us work out whether this was an OSV language (or more likely, a SVO language with the O transposed to the front for emphasis), an SOV language (very common) or a free-order language with case-markings. This may or may not matter, depending on the purpose this code is intended for (which I may not understand fully). All in all, an interesting plan, one which certainly helps focus and narrow down on some of the basic syntactic (and a few semantic) categories of English and Glosa and not a few similar languages. It looks like a nice thing to run through your mind now and then to see just what the language is doing, sort of like a computer-codable version of sentence-diagramming (remember that?) ~mark ______________________________________________________________________ >From hrick@world.std.com Tue Feb 1 17:16:48 1994 Date: Tue, 1 Feb 1994 22:16:48 -0500 From: hrick@world.std.com (Richard Harrison) Message-Id: <199402020316.AA04123@world.std.com> To: conlang@diku.dk Subject: vocabulary considerations Here's a rough sketch of an article on vocabulary-creation. Comments, suggestions, criticisms invited. 101 things to consider while creating a vocabulary * Finite or infinite: Natural languages such as English can continue adding new words to their vocabularies endlessly; there is no limit on the size of their lexicons. Many conlangs, however, have a limited-size vocabulary. In some cases, the limit is imposed by the phonetic shape of the root-words: a language based on CVC words will be limited to the number of CVC combinations possible. In other cases, such as Basic English, the limit is an arbitrary number based on the language creator's idea of how many words will be needed. * Concept mapping: Most conlang designers say there should be no synonyms in a conlang's stock of radicals. (But how far can you take this notion; for example, should the noun meaning "time," the conjunction meaning "when" and the preposition that means "during" all be derived from the same radical?) Conlang designers also often state that each word only represents one idea. In some cases it would be more accurate to say that each word represents one narrow range of closely-related concepts. After close examination we often find that a conlang's vocabulary connects concepts to words in a way that is nearly identical to the conlang creator's native language. * Compound words and derivational affixes: Conlangs that have finite-size vocabularies usually have a system of derivational affixes and allow the creation of compound words so that the limited lexicon can express a wide range of ideas. But not everyone agrees that natlang-style compounding is always a good thing; Rick Morneau has suggested that a conlang suitable for use as an interlingua in machine translation should explicitly state how the items in a compound word are related. It seems desirable that a reader or listener should be able to guess the meaning of a compound or derivative word, even if very few context clues are available. It is not hard to guess the meanings of Esperanto _preg^ejo_ (place devoted to prayer ~= church) or _piedbati_ (foot + hit ~= kick). But a newcomer to Dutton Speedwords might have a hard time puzzling out the meaning of _aqe_ (the word for "water" plus an augmentative suffix, roughly equivalent to Esperanto _akvego_) -- we might assume _aqe_ means "ocean" but it is Dutton's way of saying "steam." The number of root-words in the vocabulary seems to have a great influence on the comprehensibility of compound words. In a conlang like aUI, which only has 31 radicals, compounds are very ambiguous and it is unlikely that any two users of the language would spontaneously create the same set of compounds to express a given set of ideas. (The creator of aUI used these variations to gain insight on his patients in psychotherapy.) If we could restrict Esperanto or German or Chinese to the 2000 most frequent roots and affixes, we would probably find that speakers of these languages could create compounds and derivatives that were easy to understand, and we would probably find that most speakers were creating the same compounds to express the same ideas. Based on this hypothetical observation, we might assume that the minimum size for a basic vocabulary is somewhere between 31 and 2000 items, eh? * Hidden irregularities: Language designers sometimes state that their languages have no grammatical irregularities because, for example, nouns always form the plural by adding the same suffix and there are no irregular verbs. But if we dig a little deeper we often find that these claims are not true. For example, there might be some nouns that have a plural form ("count nouns") and some which never take the plural ("mass nouns"), as in English we never say "three electricities" or "five oatmeals." Often we find that the count noun/mass noun distinctions are the same in the conlang as in its creator's native language. And often we find that the creator has failed to mention which nouns are countable and which are mass. Even if verbs always form their various tenses in the same way, there are often huge variations in the arrangement of different verbs' "arguments." (This is one reason why I tried to reduce the number of basic verbs in Vorlin as much as possible [got it down to 3], and probably one of the reasons why Hogben used a minimal number of verboids with explicity stated argument-structures in Interglossa.) For example, English has >25 different verb patterns <1>, here are a few of them: [2A] Intransitive verbs that may be used without a complement: "We all breathe, eat and drink." [2B] Verbs used with an adverbial adjunct of distance, duration, weight etc. "The meeting lasted (for) two hours." [9] The object of the verb is a that-clause; "that" is often omitted. "I suppose (that) you'll be leaving soon." [14] The verb is followed by a direct object and a preposition and its object. This pattern is not convertible to VP12... `Explain something to somebody' cannot be converted to `*Explain somebody something.' The preposition is linked to the verb and they must be learnt together, e.g. `compare one thing TO/WITH another.' The important factor here is that any given verb can only be used in a limited number of these patterns. At the moment, I can't think of any conlangs that don't have the same complication built into them. (Loglan and its successors have taken this phenomenon to its logical extreme by making every content word a verb and giving each one a unique argument structure that must be memorized along with the word. But at least the loglans spell out in detail what the argument structures are going to be.) Bilingual conlang dictionaries seldom reveal how a verb's arguments are to be arranged, and many conlang proposals don't contain enough sample text to provide this information by example. <1> A S Hornby, Guide to Patterns and Usage in English, Oxford University Press, as quoted in Oxford Advanced Learner's English-Chinese Dictionary. ______________________________________________________________________ >From j.guy@trl.oz.au Thu Feb 3 02:47:54 1994 From: j.guy@trl.oz.au (Jacques Guy) Message-Id: <9402020447.AA16554@medici.trl.OZ.AU> Subject: Re: vocabulary considerations To: conlang@diku.dk Date: Wed, 2 Feb 1994 15:47:54 +1100 (EST) Ah, interesting stuff I will enjoy to disagree with! > > > 101 things to consider while creating a vocabulary > > * Finite or infinite: Natural languages such as English can continue > adding new words to their vocabularies endlessly; there is no limit > on the size of their lexicons. Yes there is! First, the memory of the speaker. My brain capacity has a limit. I don't know what it is, but I am sure it has. So there is a limit to what it can hold. Second, I'll grant you 120 phonemes and such a nimble tongue and vocal chords that "tQx!zz&%%%p*H" is no problem to you -- and clearly distinguished from "tQx!zz&%%%%p*H", too! Now even though I will grant you too words up to 12,000 phonemes long (will that allow you one googolplex different words? No matter if it doesn't: make them anything right up to 1,200,000 phonemes long!) there is a limit, even if you manage to stay awake long enough to utter that million-phoneme word: your lifetime -- or that of the Solar system if your name's Super Mathuselah. > > * Concept mapping: Most conlang designers say there should be no > synonyms in a conlang's stock of radicals. (But how far can you > take this notion; for example, should the noun meaning "time," the > conjunction meaning "when" and the preposition that means "during" > all be derived from the same radical?) Conlang designers also often > state that each word only represents one idea. In some cases it > would be more accurate to say that each word represents one narrow > range of closely-related concepts. I will disagree there by saying "in all cases each word represents a set of ranges of related concepts". (The set might have only one member). But I'm cheating here: it's really agreeing with you and going one (or two) better, no? I agree with all the rest of what you wrote, though! ______________________________________________________________________ >From chalmers@violet.berkeley.edu Tue Feb 1 23:35:25 1994 Date: Wed, 2 Feb 1994 07:35:25 -0800 From: chalmers@violet.berkeley.edu (John H. Chalmers Jr.) Message-Id: <199402021535.HAA15487@violet.berkeley.edu> To: conlang@diku.dk Subject: Laadan I recently bought a copy of Suzette Haden Elgin's novel "Earthsong Native Tongue III," the latest in her series about the linguist clans who communicate with visiting ET's in the context of a repressive anti-feminist US. I haven't finished it yet, so I won't attempt a review, save to say that it appears to contain little about Laadan (with an acute accent and a high pitch on first a), the conlang she created and used thematically in these books to _"express the perceptions of women"_. However an address is given for those who wish to obtain a grammar, a dictionary, and more information about the language. Send a self-addressed stamped envelope to Laadan, P.O. Box 1137, Huntsville, Arkansas, 72740 USA. I have an earlier version of the grammar and dictionary back at my home in Houston (I'm currently in Southern California). The language is rather interesting as it incorporates a number of ideas from aboriginal North American languages in a quite non-Indo-European framework. -- John ______________________________________________________________________ >From shoulson@ctr.columbia.edu Wed Feb 2 06:32:23 1994 From: shoulson@ctr.columbia.edu (Mark E. Shoulson) Received: from localhost (shoulson@localhost) by startide.ctr.columbia.edu (8.6.5/8.6.4.788743) id LAA04017; Wed, 2 Feb 1994 11:32:23 -0500 Date: Wed, 2 Feb 1994 11:32:23 -0500 Message-Id: <199402021632.LAA04017@startide.ctr.columbia.edu> To: conlang@diku.dk Subject: Laadan I bought acopy of the "First Grammar and Dictionary of L\'aadan" or whatever it's called some time ago; is there indeed a newer version? L'aadan, to me, is somewhat interesting, though I disagree with some of Elgin's premises. But that's another story. ~mark ______________________________________________________________________ >From C.J.Fine@bradford.ac.uk Wed Feb 2 17:18:09 1994 Date: Wed, 2 Feb 1994 17:18:09 GMT Message-Id: <29370.199402021718@discovery.brad.ac.uk> Received: from Colin Fine's Macintosh (colin_fine.comp.brad.ac.uk) by discovery.brad.ac.uk; Wed, 2 Feb 1994 17:18:09 GMT From: Colin Fine To: conlang@diku.dk Subject: Re: Laadan I found some of the ideas of La'adan very interesting (I bought a grammar, but I have no idea what I have done with it), but being a mere man I do not know what it is about it that is especially suited to women. I read Native Tongue and disliked it so much that I haven't read the second one (I didn't know there was a third). From the linguistic point of view, I think that she was doing just the same as some hard science SF writers used to do in the forties: give you some interesting nuggets of science (in this case linguistics) and then retreat into double-talk for the magic ray that saves the hero (in this case the magic language that saves the women). In particular she devotes a lot of space to the idea of 'encodings': new concepts for which there has not been a word before. I am convinced that insofar as a language may have potentially whorfian effects the accident of whether a concept has a word or needs to be expressed in several words is minor, compared to subtle distinctions which are expressed not in the vocabulary but in the grammar: the range and variety of grammatical categories, how aspects are handled, grammaticalisation of animacy or other hierarchies, perhaps features such as the choice of accusative vs ergative marking.... In other words, while I accept that Laadan *might* have some powerful and wonderful properties for the mental capabilities of its speakers, I believe that the arguments within the book that are apparently supposed to explain why are weak and mostly irrelevant. I also found that the depiction of the anti-feminist society got up my nose: it seemed to me to be serving the needs of polemic rather than art. This is of course a personal opinion. Colin Fine ______________________________________________________________________ >From lojbab@access.digex.net Wed Feb 2 09:39:43 1994 From: Logical Language Group Message-Id: <199402021939.AA26065@access2.digex.net> Subject: Re: Laadan To: conlang@diku.dk Date: Wed, 2 Feb 1994 14:39:43 -0500 (EST) Cc: lojbab@access.digex.net (Logical Language Group) Colin Fine writes: > insofar as a language may have potentially whorfian effects the > accident of whether a concept has a word or needs to be > expressed in several words is minor, compared to subtle > distinctions which are expressed not in the vocabulary but > in the grammar: [examples deleted] While basically agreeing with what you say, and not having read a word of Elgin, I think there may be something to be said for this notion of "encodings", if deconstructed a bit. Surely it makes little difference whether a concept is expressed in one word or many, even given a language where the notion "word" has vivid meaning (true of English, but false of German, e.g.). Nevertheless, two points come to mind: 1) Having a word for something may indicate that it is, to some degree, culturally backgrounded. As Larry Niven points out somewhere, being an "American" indicates a different mental attitude from calling yourself something that translates as "people, the people". The first case implicitly says that there exist other kinds of people, whereas the second does not: the word for "white man" in various African and Native American languages is often equal to, or related to, words for "ghost/devil/supernatural entity". I was reflecting today that Japanese has a short word (I forget it, but it has about 3-4 morae) for "non-verbal communication", which of course is a very important concept in Japanese culture. Then there is English, which wastes a monosyllable on the concept "short cylinder used in playing hockey and related games." :-) 2) If two different words exist for what another language conveys with a single word (possibly modified as needed), then the two different words may have distinct polysemy sets. In the conlang LeGuin uses in >Always Coming Home<, there are distinct monosyllabic words for "male orgasm" and "female orgasm", each with its own set of polysemous meanings. I recall that the word for "male orgasm" also means "achievement", which seems plausible; I don't recall (alas) any of the polysemous meanings for "female orgasm". So maybe "lexical Whorfianism" isn't quite the crock we enlightened folk always thought it was. -- John Cowan sharing account for now e'osai ko sarji la lojban. ______________________________________________________________________ >From ucleaar@ucl.ac.uk Wed Feb 2 19:10:32 1994 From: ucleaar Message-Id: <25303.9402021910@link-1.ts.bcc.ac.uk> To: conlang@diku.dk Subject: Re: vocabulary considerations Date: Wed, 02 Feb 94 19:10:32 +0000 Jacques to Rick H: > First, the memory of the speaker. My brain capacity has a limit. > I don't know what it is, but I am sure it has. So there is a limit > to what it can hold. You're talking about idiolect, not language. A language is a collection of features shared between idiolects. I would hold that the OED can fairly be called a dictionary of English, even though no idiolect, I would wager, contains all words listed in the OED. So I agree with Rick. And I would pose a further question: When adopting a word for a relatively novel concept, such as Boson, Kangaroo, Krill or Modem, is there anything to be gained by creating a word from preexisting morphemes? I prefer to hold to the principle that words with related meanings have related sounds, but that they needn't be analysable into component morphemes. In fact 'boson' is quite a good example of this: most names of particles end in -on, but the bos- has (to me) no independent meaning (i.e. it is a cranberry-morph). ---- And ______________________________________________________________________ >From shoulson@ctr.columbia.edu Wed Feb 2 11:08:18 1994 From: shoulson@ctr.columbia.edu (Mark E. Shoulson) Received: from localhost (shoulson@localhost) by startide.ctr.columbia.edu (8.6.5/8.6.4.788743) id QAA05589; Wed, 2 Feb 1994 16:08:18 -0500 Date: Wed, 2 Feb 1994 16:08:18 -0500 Message-Id: <199402022108.QAA05589@startide.ctr.columbia.edu> To: conlang@diku.dk Subject: Laadan My wife read Native Tongue and didn't like it either; many of her comments are similar to Colin's (though less linguistics-oriented). I see some interesting features of L'aadan, but very little that makes it so uniquely "feminine" or even all that different. Still, it has some fun parts. ~mark ______________________________________________________________________ >From lojbab@access.digex.net Wed Feb 2 11:45:45 1994 From: Logical Language Group Message-Id: <199402022145.AA02282@access2.digex.net> Subject: Re: vocabulary considerations To: conlang@diku.dk Date: Wed, 2 Feb 1994 16:45:45 -0500 (EST) And Rosta writes: > In fact 'boson' is quite a good example of this: > most names of particles end in -on, but the bos- has (to > me) no independent meaning (i.e. it is a cranberry-morph). That's because you don't belong to the relevant community of speakers. Bosons are so named because they obey "Bose-Einstein statistics", whereas fermions obey "Fermi-Dirac statistics". For those who care: Fermi-Dirac statistics apply to particles which are indistinguishable but cannot be in the same quantum state simultaneously (cannot be superposed), whereas Bose-Einstein statistics apply to particles which can be superposed. The statistics for distinguishable particles (such as billiard balls) is called "Maxwell-Boltzmann", so presumably billiard balls are maxwellons. -- John Cowan sharing account for now e'osai ko sarji la lojban. ______________________________________________________________________ >From agallagh@stars.sfsu.edu Wed Feb 2 15:03:41 1994 From: agallagh@stars.sfsu.edu (Alexis Gallagher) Message-Id: <9402030703.AA01623@stars.sfsu.edu> Subject: Re: Laadan To: conlang@diku.dk Date: Wed, 2 Feb 94 23:03:41 PST > which there has not been a word before. I am convinced that > insofar as a language may have potentially whorfian effects the > accident of whether a concept has a word or needs to be > expressed in several words is minor. . . . I am not a linguist or studying to be a linguist, but let me throw in my two bits here (though I fear, slightly, that conlang may be the wrong forum for these two bits). Somewhere in _Life of Johnson_ Boswell says how every idea or thing has a single word associated with it. Johnson, right away, points to the phrase "old age," showing how there is no single-word equivalent for it. Boswell says that certainly there is: senectus, a Latin word. At this point Johnson expatiates on how there needn't be one word in language for any particular concept, and on how two words serve just as well. I am in general inclined to agree with Johnson (and with the above quotation), but I'm not sure that the principle is absolutely true. Maybe this has been mentioned before, but it seems to me that the limits of memory would have an effect. I can know what the word 'counterpoints' means. Three syllabless are associated in my head with a fairly complex concept. Now, perhaps in a language with a smaller vocabulary (or in no language, but in thoughts themselves, if it be true that many thoughts are fundamentally non-linguistic in their representation in the brain), this same concept 'counterpoint' could be expressed as a group of simpler words, perhaps, say, as 'contrasting, balancing comparison' (though even here I seem to have failed to use noticably 'simpler' words). Perhaps I _could_ express it in simpler words, but how far can this activity be carried? Take the phrase "counterpoints the surrealism of the underlying metaphor." Each of the words in it is difficult to explain in itself, and, if each word was reduced to some compound or collection of simpler words, then I think the entire phrase would become unmanagably large--mentally unwieldy in its sheer size. It would exceed the limits of my short term memory, would thus be inexpressible and, perhaps, even unthinkable. (Short term memory may be the wrong phrase here.) I see words sort of as pegs in the mind, onto which we hang ideas. Maybe a few pegs can do the job of many but if we load them up too heavily, if we ask a few simple pegs (or even pegs just ill-suited for a cetain job) to bear the weight of an object they aren't equal or designed to carry, then they snap. We need the right peg for the right job. This is my thought. Begging pardon for the uncalled-for poeticism of that last paragraph, I wonder what the rest of you think of this notion of mine. Alexis Gallagher. ______________________________________________________________________ >From hrick@world.std.com Thu Feb 3 02:40:14 1994 Date: Thu, 3 Feb 1994 07:40:14 -0500 From: hrick@world.std.com (Richard Harrison) Message-Id: <199402031240.AA13375@world.std.com> To: conlang@diku.dk Subject: Re: Laadan agallagh@stars.sfsu.edu (Alexis Gallagher) writes: > Take the phrase "counterpoints the surrealism of the underlying > metaphor." Each of the words in it is difficult to explain in > itself, and, if each word was reduced to some compound or collection > of simpler words, then I think the entire phrase would become > unmanagably large Hmmmm, wait a minute, the main words in your example sentence already _are_ compounds. Counter- is a prefix used in other words (counter- clockwise, counterculture, etc). Real, real-ism, sur-real-ism. Under-lying. Metaphor begins with meta- (also seen in metacarpal, metalanguage, metamorphosis). (Is -phor used in any other English words? Gametophore, anaphora??) So, each of the content words in your example sentence already is a collection of simpler words; we just don't notice this immediately because we have grown accustomed to these compounds. (Perhaps if we concentrated on their component parts, rather than thinking of them as single words, it would cause an overflow in our output buffers as you suggested.) > I see words sort of as pegs in the mind, onto which we hang ideas. > Maybe a few pegs can do the job of many but if we load them up too > heavily, if we ask a few simple pegs (or even pegs just ill-suited > for a certain job) to bear the weight of an object they aren't equal > or designed to carry, then they snap. We need the right peg for the > right job. You have expressed your thought quite eloquently. Let me add that different people in different situations need different sets of pegs. People in specialized careers need specialized jargon. Vocabularies for religious concepts, foodstuffs, articles of clothing and so forth vary greatly from culture to culture. ______________________________________________________________________ >From C.J.Fine@bradford.ac.uk Thu Feb 3 13:51:07 1994 Date: Thu, 3 Feb 1994 13:51:07 GMT Message-Id: <17722.199402031351@discovery.brad.ac.uk> Received: from Colin Fine's Macintosh (colin_fine.comp.brad.ac.uk) by discovery.brad.ac.uk; Thu, 3 Feb 1994 13:51:07 GMT From: Colin Fine To: conlang@diku.dk Subject: Re: Laadan AND vocabulary I agree with John's and Alexis' points on vocabulary. Certainly the presence or absence of concise expressions for a concept, and of distinctions which in other languages are not automatically made, can have a profound impact on what gets said in the language. But I still believe that the effect is more superficial than that of deeper differences I was mentioning. An interesting example of what Alexis is saying: I have on my desk a booklet concerning Nordic Talking Books. It is in seven languages (Norwegian, Swedish, Danish, Finnish, Faroese, Icelandic and Greenlandic). The titles include: Laes Nordiske Lydbo/ger (Danish) Lesid+ norraenar hljo'd+baekur (Icelandic: d+ is the letter edh) A"a"nikirjoja pohjoismaissa (Finnish) Atuakkanik tusarnaagassianngorlugit immiussanik nunanit avannarlerneersunik atuarit (Greenlandic) The ones I understand say "Read Nordic sound-books" (I don't think the imperative 'read' is there in the Finnish). It is evident that the Greenlandic is much longer than any of the others. I deduce from the text that the Greenlandic for 'talking book' is 'atuak- tusarnaagassianngorlug- immiussa-' with some suitable collection of endings. (I guess the first word means 'book' and the third 'recorded'). This seems to me to be an example of what Alexis is talking about - this is a concept which can be expressed in Greenlandic, but not succinctly. Encodings: As I recall the discussion of encodings in Native Tongue (from a long time ago) one example given was 'the palm of the hand and the front of the forearm'. Now it may or may not be useful to have a single word for that concept (it reminds me of a discussion somewhere over whether there was an English word for the back of the knee). But I refuse to believe that the presence or absence of a word for that concept in the language is of such significance that it can have the described effect. And if Elgin is not claiming that this is a major example, then she has chosen a bad example to make her point. (I cannot remember whether it was that or another example which was described in the book as a 'significant new encoding' discovered by the young heroine). The writer's problem is of course that any concept that she can get across to the reader in English words is liable to the same criticism as I have just made. But I think she is sacrificing scientific and logical plausibility for didactic purpose. The kind of 'encoding' I might accept as plausible would be something described in terms such as 'the difference between being seen by somebody and being seen in an objective sense' This example is off the top of my head, as an example of a distinction which can be expressed in English but seems to have little or no meaning to us as English speakers - but I can conceive of a mind-set in which it is meaningful, even though I do not know what it means. Interestingly, when studying or composing Lojban I often get a sense of this sort of distinction that I am sure is real but I can't quite get my mind around. For example le ba nu mi klama = the (particular) future event(s) of I go le nu mi ba klama = the (particular) event(s) of I future go mi tadni le te djuno = I study the subject (of somebody's knowing about) mi tadni le te smadi = I study the subject (of somebody's guessing about) mi tadni le se casnu = I study the subject (of somebody's discussion) mi tadni le te jijnu = I study the subject (of somebody's intuition) Colin ______________________________________________________________________ >From chalmers@violet.berkeley.edu Thu Feb 3 23:38:48 1994 Date: Fri, 4 Feb 1994 07:38:48 -0800 From: chalmers@violet.berkeley.edu (John H. Chalmers Jr.) Message-Id: <199402041538.HAA01499@violet.berkeley.edu> To: conlang@diku.dk Subject: Laadan As well as I can remember, there is a 2nd grammar and dictionary of L/aadan as well. I sent a SASE to Huntsville, AR and will post whatever I learn. More interesting than the "encodings" perhaps are the inflectional categories of L/'aadan. They mostly differentiate degrees and type of duty, obligation ,ownership, and attitude. Whether these represent the perceptions of women is arguable -- to me they mostly describe relations of power and hence, would be equally applicable to men at the bottom of the socio-economic and political ladders. I recall from a L/'aadan newsletter I received in connection with publicity for an SF&fantasy convention which had sessions devoted to L that considerable amazement was expressed that men had devised "encodings" and transmitted them to Elgin. I tend to agree that the books are a tough read, especially for me, but I found the language interesting on its own. -- John ______________________________________________________________________ >From WEINBERG@GMUVAX.GMU.EDU Fri Feb 4 09:39:48 1994 Message-Id: <199402042242.AA24897@odin.diku.dk> Date: Fri, 4 Feb 94 14:39:48 EST From: STEVEN H. WEINBERGER To: conlang@diku.dk Subject: laadan Certainly there my be interesting linguistic relativity issues with the female language Laadan, but i have found some very instructive phonological alternations in that grammar--so instructive that i use Laadan data for a phonological exercise in my introductory phonology class at George Mason Univerisity. the students love it! --steven weinberger ______________________________________________________________________ >From chalmers@violet.berkeley.edu Fri Feb 4 11:42:01 1994 Date: Fri, 4 Feb 1994 19:42:01 -0800 From: chalmers@violet.berkeley.edu (John H. Chalmers Jr.) Message-Id: <199402050342.TAA24208@violet.berkeley.edu> To: conlang@diku.dk Subject: Laadan Steve: Could you post some examples of the use of Laadan in your introductory phonology class? I haven't looked at it for so long, I've forgotten the structure of the words. -- John ______________________________________________________________________ >From ram@eskimo.com Fri Feb 4 19:56:44 1994 Date: Sat, 5 Feb 1994 03:56:44 -0800 From: ram@eskimo.com (Rick Morneau) Message-Id: <199402051156.AA06931@eskimo.com> To: conlang@diku.dk, ram@eskimo.com Subject: Re: vocabulary considerations Don Harlow writes: > > If you intend your conlang to be used for literary purposes... > you are going to have to contend with the insistence of those using > it that there be several genuine synonyms for each concept in the > language, so that the same word doesn't have to be repeated twice > in the same paragraph or on the same page. > It's often comforting to have such synonyms, because it allows a writer (such as myself) to make something that is poorly written look like something that is well written. Nowadays, when I see myself groping for a synonym to avoid an awkward-sounding sentence or paragraph, I try to force myself to rewrite the thing from scratch, rather than use a synonym. If you've got the willpower, this ALWAYS works. However, it requires discipline which I don't always have. I see no reason to design a language to make it easy for lazy or incompetent writers, such as myself. An even better reason for avoiding synonyms, though, is that any synonyms you create are likely to represent corresponding synonyms in your native language. Other languages, of course, will not have the same synonyms. Thus, by creating such synonyms in your conlang, you'll simply be cloning your natlang. For some people, this is acceptable. For me, it is not. On a similar note, synonyms also make it easier for poets to write poetry. However, making it easier to write poetry simply cheapens the results. A conlang designer should not concern himself with poetry. Leave THAT job to the poet. In sum, a conlang designer should not coddle, spoil, pamper, baby or indulge incompetents or poets. Let them sink or float on their own. Regards, Rick *=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=* =* Rick Morneau ram@eskimo.com "Be kind to nature - =* *= Idaho Falls, Idaho, USA brake for dinosaurs." *= =*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*= ______________________________________________________________________ >From ram@eskimo.com Fri Feb 4 19:56:41 1994 Date: Sat, 5 Feb 1994 03:56:41 -0800 From: ram@eskimo.com (Rick Morneau) Message-Id: <199402051156.AA06927@eskimo.com> To: conlang@diku.dk, ram@eskimo.com Subject: Re: vocabulary considerations Howdy conlangers! First, I apologize in advance for the length of what follows. Since I now pay long-distance phone rates for every byte that I read/write via Internet, I've become very sensitive to verbosity. I promise that it won't happen often. Rick Harrison again provides some interesting food for thought, this time in an area that is closely related to one of my favorite topics: lexical semantics. Here are some of my thoughts on the points he raised: Concerning vocabulary size, compounding and derivation: First, we must make a distinction between words and root morphemes. If compounding and/or derivation is allowed, as is true with every language I'm familiar with (nat and con), then vocabulary size can be essentially infinite. Even in a system with little or no derivation (such as Chinese and Vietnamese), you can create zillions of words from compounding, even though the number of root morphemes is limited. The problem here, though, as Rick pointed out, is that you often have to metaphorically or idiomatically stretch the meanings of the component morphemes to achieve the desired result. How, for example, should we analyze English compounds such as "blueprint", "cathouse", "skyscraper" and "billboard"? Another problem surfaces if you want your compounds to be semantically precise (assuming, of course, that the basic components are semantically precise to start with). This will often mean that additional morphemes must be added to a word to indicate how the component morphemes relate to each other. For example, what is the relationship between "house" and "boat" in the word "houseboat"? What is the relationship between "house" and "maid" in the word "housemaid"? Obviously, the relationships are different. A possible solution to this problem is to create compounds by juxtaposing complete words, but keeping them separate, as is almost always done in Indonesian, and often done in English. Some English examples are "stock exchange", "money order" and "polar bear". To remove ambiguities, you will still need the additional morphemes, only this time they can be linking morphemes such as English prepositions. Swahili uses this approach for all of its compounds, and French uses it for most (French examples: "salle a manger", "eau de toilette", "film en couleurs", etc.). If you wish to use this approach, though, make sure that you have enough linking morphemes to deal with all possible semantic distinctions. Unfortunately, if you don't have a very large and expandable set of root morphemes, you'll definitely run into trouble if your goal is semantic precision. Personally, I don't like conlangs that limit the number of possible root morphemes - you never know what you're going to run into in the future. A conlang should not only give itself lots of room for expansion, but it should make it as easy as possible to implement. Another thing that should be considered is how easy it will be to learn the vocabulary. This can be best achieved by limiting the number of root morphemes. But if we limit the number of root morphemes, we run into the problems mentioned above! Actually, there is a solution to this problem. You must design your vocabulary in two steps, as follows: First, your conlang must have a powerful classificational and derivational morphology for verbs. (Other state words, such as adjectives and adverbs, will be directly derived from these verbs.) This morphology will be semantically precise. Second, root morphemes should be RE-USED with unrelated NOUN classifiers in ways that are mnemonic rather than semantically precise. I.e., the noun classifiers themselves will be semantically precise, but the root morphemes used with them (and which will be borrowed from verbs) will be mnemonic rather than semantic. To clarify the first step somewhat, let me re-post something I posted several months ago (slightly edited): ********** 1. Design a derivational morphology for your conlang that is as productive as you can possibly make it. This will almost certainly require that you mark words for part-of-speech, mark nouns for class, and mark verbs for argument structure (i.e., valency and case requirements) and voice. 2. Start with a common verb (or adjective) and decompose it into its component concepts using the above system. For example, the verb "to know" has a valency of two, the subject is a semantic patient and the object is a semantic theme. (The theme provides a focus for the state "knowledgeable". Unfocused, the state "knowledgeable" would be closer in meaning to the English words "intelligent" or "smart".) 3. The root morpheme meaning "knowledgeable/intelligent" can now undergo all the morphological derivations that are available for verbs. Some of these derivations will not have counterparts in your natlang. Many others will. For example, this SINGLE root morpheme could undergo derivation to produce the following English words: "know", "intelligent", "teach", "study", "learn", "review", "instruct", plus words derived from these words, such as "student", "intelligence", "education", etc. You will also be able to derive words to represent concepts for which English requires metaphor or periphrasis, such as "to broaden one's mind", "to keep up-to-date", etc. It is important to emphasize that ALL of these words can be derived from a SINGLE root morpheme. In other words, use a back door approach - start with a powerful derivational system, and iteratively decompose words from a natlang and apply all derivations to the resulting root morphemes. In doing so, many additional useful words will be automatically created, making it unnecessary to decompose a large fraction of the remaining natlang vocabulary. ********** Now, let me clarify the second step: Root morphemes used to create verbs can then be re-used with unrelated NOUN classificational morphemes in a way that is semantically IMPRECISE, intentionally, but which is mnemonically useful. For example, a single root morpheme would be used to create the verbs "see", "look at", "notice", etc. by attaching it to appropriate classificational affixes for verbs. These derivations would be semantically precise. The SAME root morpheme can then be used to create nouns such as "diamond" (natural substance classifier), "glass" (man-made substance classifier), "window" (man-made artifact classifier), "eye" (body-part classifier), "light" (energy classifier), and so forth. Thus, verb derivation will be semantically precise. Noun derivation, however, cannot be semantically precise without incredible complication (try to derive the word for "window" from basic primitives). So why not re-use the verb roots (which define states and actions) with noun classifiers in ways that are mnemonically significant? Finally, if you combine these two approaches with the compounding scheme mentioned earlier (using linking morphemes), you will be able to lexify any concept while absolutely minimizing the number of root morphemes in the language. Incidentally, this approach also makes it trivially easy to create a language with a self-segregating morphology. Concerning concept mapping: Again, let me save myself some typing by stealing something I wrote in that earlier post: ********** In other words, use a back door approach - start with a powerful derivational system, and iteratively decompose words from a natlang and apply all derivations to the resulting root morphemes. In doing so, many additional useful words will be automatically created, making it unnecessary to decompose a large fraction of the remaining natlang vocabulary. This approach won't guarantee that concept space will be perfectly subdivided, but it will be as close as you can get. If anyone knows of a better system, please tell us about it. Another fairly obvious advantage is that your conlang will be easier to learn, since you'll be able to create many words from a small number of basic morphemes. Ad hoc borrowings from natlangs will be minimized. Also, such a rigorous approach to word design has some interesting consequences that may not be immediately obvious. If you use this kind of approach, you'll find that many of the words you create have close (but not quite exact) counterparts in your native language. However, this lack of precise overlap is exactly what you ALWAYS experience whenever you study a different language. In fact, it is this aspect of vocabulary design that seems to frustrate so many conlangers, who feel that they must capture all of the subtleties of their native language. In doing so, they merely end up creating a clone of the vocabulary of their natlang. The result is inherently biased, semantically imprecise, and difficult to learn for speakers of other natlangs. It is extremely important to keep in mind that words from different languages that are essentially equivalent in meaning RARELY overlap precisely. Fortunately, all of this does NOT mean that your conlang will lack subtlety. In fact, with a powerful and semantically precise derivational morphology, your conlang can capture a great deal of subtlety, and can go considerably beyond any natural language. The only difference is that, unlike a natural language, the subtleties will be predictable rather than idiosyncratic, and the results will be eminently neutral. So, do you want to create a clone of an existing vocabulary? Or do you want to maximize the neutrality and ease-of-learning of the vocabulary of your conlang? You can't have it both ways. ********** Concerning hidden irregularities: A classificational system automatically solves all count/mass/group problems, since the classification will indicate the basic nature of the entity represented by the noun. Other derivational morphemes (let's call them "class-changing morphemes") can then be used to convert the basic interpretation into one of the others. For example, from the basic substance "glass", we can derive the instance of it: "a glass item". From the basic animal "sheep", we can derive its group meaning, "flock", and its mass meaning, "mutton". Each basic classifier would have a default use depending on the nature of the classifier. Further derivation would be used to create non-default forms. With this approach, it would not even be possible to copy the idiosyncratic interpretations from a natural language, since the classificational system would eliminate all such idiosyncrasy. All of the problems of verbal argument structure are solved in a classificational system. My essay goes into a lot of detail on this point, so I won't say much here. Basically, though, verbs are created by combining a root morpheme that indicates a state or action with a classifier which indicates the verb's argument structure. For example, the following verbs are formed from the same root morpheme, but with different verbal classifiers that indicate the verb's argument structure: to teach (someone): subject is agent, object is patient to teach (something): subject is agent, object is theme to learn: subject is patient, object is theme to study: subject is both agent and patient, object is theme As illustration, the semantics of the English verb "to teach someone something" can be paraphrased as: 'agent' causes 'patient' to undergo a change of state from less knowledgeable to more knowledgeable about 'theme'. You will also need to make distinctions between steady state and change of state. The above examples all indicate changes of state (i.e., the 'patient' gains in knowledge). Some steady-state counterparts, formed from the same root morpheme, would be: to know: subject is patient, object is theme to be knowledgeable or smart: subject is patient, no object to review (in the sense "keep oneself up-to-date"): subject is both agent and patient, object is theme You will also need an action classifier, which would indicate an ATTEMPT to achieve a change of state, but with no indication of success or failure. For example, the root morpheme for the above examples could be combined with an action classifier to create the verb "to instruct". Thus, the verb classifier indicates the verb's argument structure, and allows creation of related verbs from the same root morpheme, verbs that almost always require separate morphemes in English. Finally, if your conlang has a comprehensive system for grammatical voice, even more words can be derived from the same morpheme. For example, if your language has an inverse voice (English does not), you could derive the verbs "to own" and "to belong to" from the same root morpheme. Ditto for pairs such as "parent/child", "doctor/patient", "employer/employee", "left/right", "above/below", "give/obtain", "send/receive", etc. Note that these are not opposites! They are inverses (also called converses). Many other words can also be derived from the same roots if your conlang implements other voice transformations such as middle, anti-passive, instrumental, etc. You can save an awful lot of morphemes if you do it right. And even though English doesn't do it this way, there are many other natural languages that do. So there's nothing inherently unnatural about this kind of system. It's almost certain, though, that no SINGLE natural language has such a comprehensive and regular system. Finally, for those among you who want a Euroclone, I'm sorry, but I have nothing to offer you. Besides, I doubt if any of you even got this far. :-) (BTW, how to do all of the above and much more is the topic of the essay on Lexical Semantics that I am currently working on, and which was recently mentioned on this list. It goes into much, much more detail than what I've provided here. I will let the list know when it is done.) Ciao for nao! Rick *=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=* =* Rick Morneau ram@eskimo.com "Be kind to nature - =* *= Idaho Falls, Idaho, USA brake for dinosaurs." *= =*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*= ______________________________________________________________________ >From EZ-as-pi@cup.portal.com Fri Feb 4 20:53:41 1994 From: EZ-as-pi@cup.portal.com To: conlang@diku.dk, ez-as-pi@cup.portal.com Subject: Re: vocabulary considerations Lines: 44 Date: Sat, 5 Feb 94 04:53:41 PST Message-Id: <9402050453.1.11158@cup.portal.com> Rick Morneau recently repeated some earlier-posted recipes for vocabulary formation. I think they are interesting and I might even be in agreement with a fair proportion of the ideas he expressed. I would, however, like to say that the test of the practicality of such a scheme is someone's actually attempting to follow such a plan. Some months ago, I put forth a scheme for building a syntactic system, essentially a reversal of the system of Japanese. A few people joined me in the project, though it ultimately foundered because the group was small enough that the departure of just a few from the project left us with the un- fortunate situation that there was no way of achieving consensus when two people disagreed on an idea, unless the third was allowed by himself to make the decision. When there are three active people, the person who cares the least about some point at issue gets to be the decision maker. Whether it is one person, or a group, unless somebody (or -bodies!) decides to follow up Rick's ideas, we don't really know that his somewhat vague and abstract proposals can in fact be followed. While, after the Voksigid experience, I am rather gun-shy about offering to be part of a group to develop Rick's ideas, I would appreciate it if anyone who does could keep me informed about any productive follow up to his note. Bruce R. Gilson ______________________________________________________________________ >From EZ-as-pi@cup.portal.com Fri Feb 4 21:10:55 1994 From: EZ-as-pi@cup.portal.com Received: from localhost (pccop@localhost) by hobo.corp.portal.com (8.6.4/1.16) id FAA16557 for conlang@diku.dk; Sat, 5 Feb 1994 05:10:56 -0800 To: conlang@diku.dk, ez-as-pi@cup.portal.com Subject: Re: vocabulary considerations: verb patterning Lines: 36 Date: Sat, 5 Feb 94 05:10:55 PST Message-Id: <9402050510.1.11158@cup.portal.com> Rick Harrison remarked about the vast variety of verb patterns in English. I think that here I'd like to put in a plug for the approach we followed in Voksigid: every verb is in fact an "impersonal" in structure. Explicit use of prepositional phrases, with well-defined meanings for the preposi- tions, related all nouns to the main verb of the sentence. Therefore, the word order was rather flexible in that the individual phrases could occur in any sequence that emphasis could suggest: Voksigid was not really VSO or VOS, but merely verb- first, with "subject" and "object" not being valid concepts of the language; instead we could consider the active doer of the verb in question, the recipient, the undergoer, etc. merely as being defined by these prepositions just as, in English, locatives can be marked by "in" or instrimentals by "with" or "by means of"; and the dif- fering ways "subjects" or "objects" relate to the verb can be expressed by using different prepositions. I admit tht wegot into some difficulties. I have always felt that in the case of verbs of sensation, for example, the logical subject corresponds to the English object and vice versa, but not all of us agreed. But this could have been worked out if we had more time and people on the project. Bruce R. Gilson ______________________________________________________________________ >From EZ-as-pi@cup.portal.com Fri Feb 4 21:25:13 1994 From: EZ-as-pi@cup.portal.com Received: from localhost (pccop@localhost) by hobo.corp.portal.com (8.6.4/1.16) id FAA19490 for conlang@diku.dk; Sat, 5 Feb 1994 05:25:16 -0800 To: conlang@diku.dk, ez-as-pi@cup.portal.com Subject: Re: vocabulary considerations Lines: 18 Date: Sat, 5 Feb 94 05:25:13 PST Message-Id: <9402050525.1.11158@cup.portal.com> And Rosta wrote: > ... In fact 'boson' is quite a good example of this: >most names of particles end in -on, but the bos- has (to >me) no independent meaning (i.e. it is a cranberry-morph). To And, this is so. To me the associa- tion of boson with Bose is there, as I am familiar with the distinction between bosons (which "obey Bose- Einstein statistics") and fermions (which "obey Fermi-Dirac statistics"). A lot of the associations that one makes between related words depend on _knowing_ the related words. Bruce R. Gilson ______________________________________________________________________ >From WEINBERG@GMUVAX.GMU.EDU Sun Feb 6 17:03:50 1994 Message-Id: <199402070304.AA02706@odin.diku.dk> Date: Sun, 6 Feb 94 22:03:50 EST From: STEVEN H. WEINBERGER To: conlang@diku.dk Subject: Laadan as a phonology problem Engl 690 This is a subset of the Laadan examples i use in my introductory phonology class at George Mason University. Ladan is a language constructed by a woman for women, for the specific purpose ofexpressing the perceptions of women. This exercise will not address the whorfio-socio-political aspects of Ladan. It will instead focus on the phonology of this constructed language. describe the epenthesis and deletion phenomena in the following data: A. bII hal ra omId wa the horse doesn't work. DECL work NEG horse TRUTH bII mEhal ra omId wa the horses don't work. DECL work(pl.) NEG horse TRUTH bII aja mahIna wa the flower is beautiful. DECL beautiful flower TRUTH bII mEhaja mahIna wa the flowers are beautiful. DECL beautiful(pl.) flower TRUTH (Answer: we asssume that the plural prefix is /mE/. /h/ is added to keep vowels apart.) B. hohazh airport hohazhEdI airport (goal) hoth place hothEdI place (goal) marI island marIdI island (goal) Eba spouse EbadI spouse (goal) (Answer: the "goal" suffix is /EdI/. the initial vowel in the suffix is deleted if the root ends in a vowel.) ______________________________________________________________________ >From hrick@world.std.com Mon Feb 7 02:27:23 1994 Date: Mon, 7 Feb 1994 07:27:23 -0500 From: hrick@world.std.com (Rick Harrison) Message-Id: <199402071227.AA09958@world.std.com> To: conlang@diku.dk Subject: Re: vocabulary considerations pa ze Morno kwi Rihk (Rick Morneau) shu: > Root morphemes used to create verbs can then be re-used with unrelated > NOUN classificational morphemes in a way that is semantically IMPRECISE, > intentionally, but which is mnemonically useful. Well, I was cheering for you up to this point. This suggestion is a bit diappointing. Why be so precise and predictable in the rest of your proposal, and then take such a fuzzy approach to noun creation? This seems terribly inconsistent. > For example, a single root morpheme would be used to create the verbs > "see", "look at", "notice", etc. by attaching it to appropriate > classificational affixes for verbs. These derivations would be semantically > precise. The SAME root morpheme can then be used to create nouns such as ... > "glass" (man-made substance classifier), "window" (man-made artifact Why not "eyeglasses" rather than "window"? Really, this blurry method of noun creation opens the door to an endless stream of quibbles and misinterpretations. > Noun derivation, however, cannot be semantically precise without > incredible complication (try to derive the word for "window" from > basic primitives). English "window" comes from Old Norse vindr + augu (wind-eye). In other words, it already _is_ derived from basic primitives. In Chinese, "window" _is_ a basic primitive, chuang1.* If excruciating precision is our goal, we might creating a compound meaning "opening in wall for:the:purpose:of passage done:by air and/or light." But this would be rather clumsy in your proposed system, as "light" is already "(energy somehow associated with) seeing" and "air" would presumably be "(natural substance somewhow associated with) breathing." (I shudder to think how you would express "wall"! Vertical planar man-made artifact for dividing interior spaces?) (By the way, isn't "man-made artifact" redundant?) I'm looking forward to your essay. Hopefully it will include a demo conlang (more than just a few words) so that the audience can test the viability of your heretical proposals. --- * You propose atomizing meanings to the point that they are no longer recognizable, much as a common object such as a hair or a grain of salt becomes unrecognizable when viewed through a powerful microscope. Graphing the semantic domain in a way that it so terribly different from major natlangs would appear to increase the difficulty of learning. A conlang built as you propose might have a relatively small stock of roots, but this would be offset by having to learn the relatively bloated inventory of derivational affixes and the rules that control each, plus having to memorize the unpredictable meanings of all those nouns. ______________________________________________________________________ >From ram@eskimo.com Wed Feb 9 19:19:43 1994 Date: Thu, 10 Feb 1994 03:19:43 -0800 From: ram@eskimo.com (Rick Morneau) Message-Id: <199402101119.AA11871@eskimo.com> To: conlang@diku.dk, ram@eskimo.com Subject: Re: vocabulary considerations Howdy conlangers! Rick Harrison writes: > > pa ze Morno kwi Rihk (Rick Morneau) shu: > Could you identify the above language(s) for me? I wrote: > > Root morphemes used to create verbs can then be re-used with unrelated > NOUN classificational morphemes in a way that is semantically IMPRECISE, > intentionally, but which is mnemonically useful. > Rick Harrison responds: > > Well, I was cheering for you up to this point. This suggestion is a bit > diappointing. Why be so precise and predictable in the rest of your > proposal, and then take such a fuzzy approach to noun creation? This > seems terribly inconsistent. > I'm sorry you're disappointed. Perhaps you misunderstand what I'm trying to achieve. Keep in mind that I'm talking about a CLASSIFICATIONAL language where classifying morphemes are used in both verb and noun formation. Since there is no way to use verbal roots with noun classifiers, and vice versa, in a way that is semantically precise, you can either create a completely different set of root morphemes for nouns, or you can re-use the verb roots for their mnemonic value. Thus, for nouns, the combination of root+classifier becomes a de facto new root, even though it has the morphology of root+classifier. There is nothing "fuzzy" about it as long as you keep in mind that it's just a mnemonic aid. To me, it seems like a great way to re-use roots that would otherwise be underutilized. Me again: > > "glass" (man-made substance classifier), "window" (man-made artifact > Rick Harrison again: > > Why not "eyeglasses" rather than "window"? Really, this blurry method of > noun creation opens the door to an endless stream of quibbles and > misinterpretations ... > > ... English "window" comes from Old Norse vindr + augu (wind-eye). In > other words, it already _is_ derived from basic primitives. In Chinese, > "window" _is_ a basic primitive, chuang1.* > Sorry, but I don't understand how your Norse and Chinese examples are relevent. If anything, they seem to support my point. Most complex nominals used in natural languages are not semantically precise - they simply provide clues. What I'm suggesting is something akin to "blurry" English words such as "whitefish", "highland", "seahorse", etc., only the noun classifiers themselves would be more generic, but would have semantically precise definitions. Thus, what I proposed is actually much closer to what is done in Bantu languages such as Swahili, since it is morphological rather than lexical. In essense, I am suggesting that you use semantic precision only when it is practical. Re-use root morphemes as mnemonic aids when semantic precision is not practical. The alternative is to create many hundreds (perhaps thousands) of additional root morphemes which will have to be learned by the student. Rick Harrison again: > > (I shudder to think how you would express "wall"! Vertical planar > man-made artifact for dividing interior spaces?) > Of course not. I hoped that my noun examples would make it obvious that my goal is to AVOID semantic decomposition of basic nouns, since it could never succeed without resulting in incredibly long words, as your "wall" example indicates. I don't understand how you could have interpreted my examples in this way. Rick Harrison again: > > * You propose atomizing meanings to the point that they are no longer > recognizable, much as a common object such as a hair or a grain of > salt becomes unrecognizable when viewed through a powerful microscope. > Absolutely not. Again, I don't understand how you could come to a conclusion like this from what I wrote. (Perhaps I should not have tried to summarize a very long essay in just a few paragraphs.) Verb design is "atomized" only to the extent that argument structure and basic verbal nature (steady-state, change-of-state or action) are specified by the classifier. I thought my examples made this obvious. Thus, a single root morpheme plus one classifier would result in the verb "to teach". The same root morpheme plus a different classifier would result in the verb "to study". And so forth. Is this "atomizing"? If so, I advise you to avoid languages like Arabic, Turkish, Swahili, Tamil, Hungarian, Quechua, Japanese, Indonesian, Hindi, Korean, ad nauseam. Even languages closely related to English (such as French and German) make distinctions in verbal morphology between the agentive "He opened the door" and the stative, non-agentive "The door opened". The non- European languages in the above list make even more distinctions. In fact, all of the derivational morphology in my essay has counterparts in natural languages. The only difference is that (to my knowledge) no single language makes ALL of the distinctions I make, nor do they implement them with total regularity. And as for nouns, the combination of verbal root plus noun classifier becomes a de facto noun root. The classifier has a precisely defined meaning, while the verbal root simply makes a mnemonic contribution to the final meaning. The word "atomizing" simply does not apply, unless you consider ANY type of classification as "atomizing". Rick Harrison again: > > Graphing the semantic domain in a way that it so terribly different > from major natlangs would appear to increase the difficulty of learning. > Both noun and verb classifiers are used, in effect, to create most of the vocabularies of languages such as Swahili and Arabic. Also, many oriental languages make heavy use of lexical classification schemes. I admit, though, that my approach has little in common with most European languages. (Which I consider a definite advantage. Euroclones are a dime a dozen.) Even so, there is nothing "terribly different" about my scheme. English creates many complex nominals this way (eg. "cutworm", "white water", "red ant", etc.). My approach, though, uses noun classifiers that are slightly more generic than "worm", "water" and "ant". In effect, it is much more similar to Bantu languages of Africa or several aboriginal languages of Australia. These languages, though, are at the opposite extreme from English, since their classifiers are even vaguer than what I propose. Thus, my ideas fit in quite snugly between the opposite poles of classificational possibility. Difficult??? Adding regularity to word design will make it easier, not more difficult. Is Esperanto more difficult because it's inflectional system is perfectly regular? Of course not. Just because perfect regularity in a natlang is extremely rare does not mean that we should avoid it in the construction of a conlang. Or are you saying that it's okay to have regularity in syntax and inflectional morphology, but that it's NOT okay to have regularity in derivational morphology or lexical semantics? I suggest that most conlangs are irregular in derivational morphology and lexical semantics because their designers are not aware that such regularity is even possible. Rick Harrison again: > > A conlang built as you propose might have a relatively small stock > of roots, but this would be offset by having to learn the relatively > bloated inventory of derivational affixes and the rules that control > each, plus having to memorize the unpredictable meanings of all those > nouns. > Instead of being forced to learn thousands of unique-but-related verbs, I would rather learn about one-tenth as many, plus a few dozen classifiers and a few perfectly regular rules that apply without exception. As for nouns, mnemonic aids make them easier to learn - their meanings are unpredictable only if you fool yourself into thinking that they SHOULD BE predictable. Rick, I think (hope?) that there are two reasons why you have difficulty with my post. First, you raised a topic that I've given a lot of thought to, and I tried to summarize a large quantity of material that I've written on the topic in just a few paragraphs. Misunderstanding was inevitable. Second, a classificational language may not hold much appeal for you. If so, I'm sure you're not alone. I choose this approach because it has several advantages. First, and least important, it makes word design fast and easy. Second, it makes learning the language easier. Third, it is totally neutral - no one will accuse you of cloning your native language. Yet nothing in my approach is unnatural - every aspect of it has counterparts in some natural languages. Fourth, and most importantly, is that a powerful classificational and derivational system forces the conlang designer to be systematic. If done properly, it will prevent the adoption of ad hoc solutions to design problems. Aaaiiieeeyaaah! That fourth point is SO important, that I want to repeat it. But I won't. :-) I also believe that the result will have more esthetic appeal to a larger number of people of varied backgrounds. A conlang with a large contribution from European languages may appeal to Europeans, but it will probably not be as appealing to non-Europeans. Regards, Rick *=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=* =* Rick Morneau ram@eskimo.com "All kings is rapscallions" *= *= Denizen of Idaho, USA --Mark Twain =* =*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*= ______________________________________________________________________ >From ucleaar@ucl.ac.uk Thu Feb 10 21:20:51 1994 From: ucleaar Message-Id: <63661.9402102120@link-1.ts.bcc.ac.uk> To: conlang@diku.dk Subject: Re: vocabulary considerations Date: Thu, 10 Feb 94 21:20:51 +0000 Rick M: > Rick Harrison writes: > > > > pa ze Morno kwi Rihk (Rick Morneau) shu: > > > Could you identify the above language(s) for me? It's Vorlin, I reckon. I seem to remember Rick once explaining _ze_ and _kwi_ as Vorlin markers of personal and family names. It's a shame we don't hear much about Vorlin on Conlang these days. ---- And ______________________________________________________________________ >From fritz@rodin.wustl.edu Thu Feb 10 20:58:48 1994 Date: Fri, 11 Feb 94 02:58:48 CST From: fritz@rodin.wustl.edu (Fritz Lehmann) Message-Id: <9402110858.AA05773@rodin.wustl.edu> To: conlang@diku.dk Subject: Intro/AI/compound-nouns/concept-systems Dear Conlangers: 10 Feb 1994 conlang@diku.dk I'm new to this list. By way of introduction, I'm working in Artificial Intelligence, particularly semantic networks and ontological systems (conceptual hierarchies) for the real world. I see the inventors of a-priori philosophic planned languages (and pasigraphic languages) as independent thinkers who deal with many deep conceptual issues needed for pragmatic communication. Their ideas may be important sources for artificial intelligence, and conceptual systems developed in artificial intelligence may be useful sources for systems of primitives in planned languages. Included at the end of this message is a list which I've compiled of all the potentially useful concept systems I know about from Aristotle up to the latest computerized taxonomies. I sent a copy to Rick Harrison already. Notice that several constructed languages appear on the list. Maybe there are others (those with carefully thought out conceptual primitives, etc., rather than the volapu-esperantoids) which I don't know about which sould be on the list --- if so, please let me know. I believe concepts form networks (semantic networks) rather than strings, so planned languages which can be spoken or written linearly are stuck with the job of converting a network into a string. This is usually done with syntax trees (and pronoun co- reference where cycles occur). Rick Morneau replying to Rick Harrison defended fine-grained conceptual compositionality for some noun combinations. I support this, but I too am bothered by noun compounds in which the relation between the primitive noun elements is "assumed". A "mnemonic" composition isn't good enough, in my books. A combination of "alligator" and "shoes" should differ from the one of "horse" and "shoes" -- and "olive"+"oil" should differ from "baby"+"oil". What's missing is usually the relevant case-like- relation (like USED-FOR or PART-OF or AGENT-OF). It seems to me that there are few enough of these basic relations that they could be very compactly encoded (say with vowel-dipthongs). If oil is BOR, OLIVE is BAK, and baby is BIK, I think BORaiBAK and BORiuBIK are not much more burdensome on the speaker than BORBAK and BORBIK, but vastly better if "ai" is "DERIVED-FROM" and "iu" is something like "USED-FOR". Case-relations are not just separate entitites -- they occur in a hierarchy of such relations. (See A.F. Parker-Rhodes "Inferential Semantics", Harvester/Humanities Press, 1978, Chapter XI.) Some case-relationss are more general than (i.e. they subsume) other more specific ones, so the case- mark in compound nouns ned not refer to a too-specific case- relation. Because I'm so new I don't even know what language they are talking about -- presumably one being devised by Rick Morneau. (It sounds very interesting and I'd like to know where I could see a description.) The problem of the "unspecified relation" plagues the compound predicates and the predicate juxtapositions in Loglan/Lojban too, as Paul Doudna and others pointed out some years ago. Yours truly, Fritz Lehmann 4282 Sandburg, Irvine, CA 92715 714-733-0566 fritz@rodin.wustl.edu ==================================================================== LIST APPENDED: [Version 4: Thanks to Piet-Hein Speel, Paul Doudna, Kurt Godden, Nick Youd, Jim Fulton and especially to Dan Fass for COPIOUS new material. Changes: RICHENS/MASTERMAN/WILKS is now divided into: RICHENS 100 MINIMALS (he invented "semantic nets" in 1956), MASTERMAN'S SEMANTIC LATTICES, PREFERENCE SEMANTICS PRIMITIVES (Wilks), and FASS'S GENUS CLASSIFICATION. The RELATIONAL LEXICON HIERARCHY becomes EVENS/NUTTER LEXICAL RELATION HIERARCHY. New total= 147 concept-systems. FL 11/12/93] =================================================================== CONCEPT-SYSTEMS CATALOGUE Fritz Lehmann 4282 Sandburg Way, Irvine, CA 92715 USA (714)733-0566 [occasionally accessed email: fritz@rodin.wustl.edu] Version 4 of: November 1993 This is to be an informal catalogue of existing concept catalogues, taxonomies and hierarchies (including high level "ontologies") for possible use in knowledge representation, artificial intelligence, simulation, and database integration. Anybody can contribute (and be acknowledged). Each concept system is to be described (in a page or less) with some references and other information. I hope to be inclusive, with emphasis on potentially machine-readable/usable concept (and relation) hierarchies. Some people think there is ONE concept system for the true structure of the world. Others like me think pragmatic concerns (subjective, mission-determined, or socially agreed-upon) may dictate different structures. Most "ontologies" have large areas of near-agreement on concepts like time, space, individuals, properties, etc. Technical thesauri deal with more specific subject areas like accounting, subfields of medicine, or plumbing fixtures. Philosophical concepts are necessary but controversial; some concepts like "check-stub" are quite uncontroversial. The list is now intentionally a grab-bag. It ranges from universal to fairly problem-specific, informal to formal. Formalized or not, two aspects of every system are: its purely mathematical (order) structure, and the meanings of its components. Notation or language is incidental to both. The page ordering, for now, is vaguely chronological. A concept in one system may differ entirely from a concept with the same name in another system. Please let me know of ANY OTHER concept-systems you know about. ------------------------------------------------- [I started this list Nov. 20, 1992 for the "PEIRCE project" (a cooperative international implementation of a Conceptual Graphs inferential database processing sytem, initiated by Gerard Ellis and Robert Levinson), with a list of 84 systems beginning with Aristotle's. CODES: (i+)= I now have a little information; (i-)= I have almost no information; (I)= I have enough information; (p+)= Page written; (p-)= Page not written; (nr)= Need references; (nd)= Need descriptive documents; (c=)= Contributed information, besides me.] ARISTOTLE'S CATEGORIES (i+,p-,nr) LLULL'S ARS MAGNA (I,p+) LEIBNIZ'S ARS COMBINATORIA & CHAR. UNIVERSALIS (I,p+) LODWYCK'S COMMON WRITING (i+,p-,nd) DALGARNO'S ARS SIGNORUM (i-,p-,nr,nd) WILKINS'S PHILOSOPHICAL LANGUAGE (I,p-) LINNAEUS BIOLOGICAL TAXONOMY (i+,p-,nr) (ANONYMOUS, c. 1830) UNIVERSAL CHARACTER (I,p-) CAVE BECK (i-,p-,nr,nd) KANT'S CATEGORIES (i+,p-,nr,nd) ROGET'S THESAURUS (I,p-) PEIRCE'S CATEGORIES (I,p-,nr) BOLZANO (i+,p-,nr,nd) MEINONG (i-,p-,nr,nd) BRADLEY (i-,p-,nr,nd) DEWEY DECIMAL, BLISS, UDC & LIBRARY OF CONGRESS (I,p-) HUSSERL'S ONTOLOGY (i+,p-,nr,nd) PRINCIPIA MATHEMATICA (I,p-,nd) WHITEHEAD'S PROCESS THEORY (i+,p-,nd) LIESNIEWSKI'S MEREOLOGY & SO-CALLED ONTOLOGY (i+,p-,nr,nd) BASIC ENGLISH (I,p-,c=Fass,Thorson) SEMANTOGRAPHY/BLISSYMBOLICS SYMBOLS (I,p-) RICHENS'S 100 "SEMANTIC NET" MINIMALS (i+,p-,nr,nd,c=Fass) CECCATO'S CORRELATION NET PRIMITIVES (i-,p-,nr,nd) MASTERMAN'S SEMANTIC LATTICES (i+,p-,nr,nd) LINCOS INTERPLANETARY LANGUAGE (I,p-,c=Godden) R.M. MARTIN'S SEMIOTIC PRIMITIVES (i+,p-,nr,nd) COLON FACETED LIBRARY CLASSIFICATION (i+,p-,nd) THE SYNOPTICON (FOR ENCYC. BRIT. GT BOOKS) (I,p-,c=Salsman) DEEP CASE SYSTEMS (i+,p-,nd) LOGLAN/LOJBAN SEMANTIC PRIMITIVE WORD ROOTS (I,p-) INGARDEN'S ARISTOTLE REVISION (i-,p-,nd) LAFFAL'S CONCEPT DICTIONARY (I,p-) LEECH'S SEMANTICS (i-,p-,nr,nd,c=Fass) SCHANK'S CONCEPTUAL DEPENDENCY THEORY (I,p-) ACM COMPUTER SCIENCE CLASSIFICATION (i-,p-,nr,nd) SHUM "SPIRITUAL" NETWORKS (i-,p-,nr,nd) ANTHROPOLOGICAL CLASSIFICATIONS (i-,p-,nr,nd) PROPAEDIA OF ENCYLOPAEDIA BRITANNICA (I,p-,c=Van Roy) WEBER RUSSELL'S CATEGORIES OF NOMINALS (i-,p-,nd,c=Fass) PARKER-RHODES' INFERENTIAL SEMANTICS LATTICES (I,p-) WIERZBICKA'S LINGUA MENTALIS (i+,p-,nd) SCHEELE'S ORDNUNG DES WISSENS (i-,p-,nr,nd,c=ISKO) PATENT CLASSIFICATION SYSTEMS (i-,p-,nr,nd,c=ISKO) NTIS/DOD/COSATI CLASSIFICATION SCHEME (i-,p-,nr,nd,c=ISKO) UNESCO THESAURUS (i-,p-,nr,nd,c=ISKO) BROAD SYSTEM OF ORDERING (i-,p-,nr,nd,c=ISKO) RUSSIAN MISON "RUBRICATOR" CLASS. CODES (i-,p-,nr,nd,c=ISKO) BHATTACHARYA'S CLASSAURUS (i-,p-,nr,nd) AUSTIN'S PRECIS CONCEPT ANALYSIS & INDEXING (i-,p-,nr,nd) KAMP'S DISCOURSE REPRESENTATION STRUCTURES (i-,p-,nr,nd) PREFERENCE SEMANTICS PRIMITIVES (I,p-,c=Wilks,Fass) MILLER/JOHNSON-LAIRD PRIMITIVES (i+,p-,nr,nd,c=Fass) HAYES'S NAIVE PHYSICS (i+,p-,nr,nd) LEHNERT'S OBJECT PRIMITIVES (i+,p-,nd,c=Fass) SCHANK/CARBONELL SOCIAL/POLITICAL ACTS (i+,p-,nd,c=Fass) EXPLANATORY-COMBINATORY DICTIONARY (MEANING-TEXT) (i+,p-,nd) MeSH - MEDICAL SUBJECT HEADINGS THESAURUS (i-,p-,nd) JOLLEY'S HOLOTHEME (I,p-) ZARRI'S RESEDA ONTOLOGY (I,p-) LENAT'S AM/EURISKO MATH CATEGORIES (I,p-) SOWA'S CONCEPTUAL GRAPHS PRIMITIVES (I,p-) BARWISE/PERRY/DEVLIN SITUATION SEMANTICS (I,p-) SCHUBERTIAN ("ECO") SUBHIERARCHIES (I,p-) FICTION CLASSIFICATION SCHEMES (i+,p-,nr,nd,c=ISKO) WAHLIN'S T.I.M, MANUFACTURING FACETS (i+,p+,nd,c=ISKO) CITIZENS ADVICE BUREAU CLASSIFICATION UK (i-,p-,nr,nd,c=ISKO) QUALITATIVE PHYSICS PRIMITIVES (i-,p-,nr,nd) SMITH-MULLIGAN ONTOLOGY (i+,p-,nr,nd) SIMONS' PART SYSTEM (i-,p-,nr,nd) SMALLTALK DATA TYPE TREE (i-,p-,nr,nd) OBJECTIVE-C (NeXTSTEP) DATA TYPE TREE (i-,p-,nr,nd) BOOCH/RATIONAL OBJECTS HIERARCHIES (i-,p-,nr,nd) WUESTER'S GENERAL THEORY OF TERMINOLOGY (i+,p+,nd,c=ISKO) KAB GERMAN LIBRARY CLASSES (K.LEHMANN) (i-,p-,nd,c=ISKO) GRAESSER'S MULTIPLE CONCEPT HIERARCHIES (i+,p-,nd) DAHLGREN/McDOWELL NAIVE SEMANTICS (i+,p-,c=Youd) LONGMAN DICTIONARY CODINGS (INCL. SLATOR) (i+,p-,nr,nd) LONGMAN LEXICON (THESAURUS) (i-,p-,nr,nd) BURGER'S THE WORDTREE (I,p-) FASS'S GENUS CLASSIFICATION (I,p-,nd,c=Fass) PENMAN UPPER MODEL (I,p-,c=Hovy) ICONCLASS ART SUBJECT CLASSIFICATION (i+,p-) SPARCK JONES/BOGURAEV DEEP CASE LIST (I,p-) COOK ONTOLOGY (i-,p-,nr,nd) DIXON ONTOLOGY (i-,p-,nr,nd) VARIOUS WILLE-STYLE FORMAL CONCEPT LATTICES (i+,p-,nr,nd) RUSSIAN MERONOMY/TAXONOMY (SHREIDER ET AL.) (i-,p-,nr,nd) SOMERS'S CASE GRID (i+,p-,nd) CHAFFIN'S RELATION HIERARCHY (i+,p-,nd) LENAT/GUHA CYC PROJECT (i+,p-,nd) EPSTEIN AM-BASED GRAPH THEORY HIERARCHY (i-,p-,nd) MARTY'S SEMIOTIC LATTICES (i-,p-,nr,nd) GEOGRAPHIC DATABASE CONCEPTS (i-,p-,nr,nd) MACKWORTH/REITER MAPSEE GEOGRAPHIC MAP AXIOMS (i+.p-,nd) IRDS DATABASE CATEGORIES (i-,p-,nr,nd) HOBBS' COMMONSENSE ONTOLOGY (I,p-) GORANSON SYMMETRIES (i-,p-,nr,nd) PANSYSTEMS PHILOSOPHICAL LOGIC (CHINESE) (i+,p-,nd) GIUNCHILIA'S ITALIAN PREPOSITIONS (i-,p-,nr,nd) LILOG ONTOLOGY (i-,p-,nr,nd,c=Speel) VELARDI'S SEMANTIC LEXICON FOR ITALIAN (i-,p-,nr,nd) ONTEK, INC.'S ONTOLOGY (i-,p-,nd) LAKOFF'S CATEGORIES (i+,p-,nd) HUHNS/STEPHENS RELATION FEATURES (i+,p-,nd) WORDNET (I,p-,c=Consortium) EDR CONCEPT DICTIONARY (JAPAN) (i-,p-,nr,nd) ONTOS CONCEPT HIERARCHY (i-,p-,nr,nd) DOUDNA QUANTIFIER RHOMBIDODECAHEDRON (I,p-,c=Doudna) SCRAMBLED ROGET (5th Ed.) (i-,p-,nr,nd) NIRENBURG'S DIONYSUS ONTOLOGY (i-,p-,nr,nd) SCHUBERT/HWANG EPISODIC LOGIC CATEGORIES (i+,p-,nd) UNITRAN-LCS (i-,p-,nr,nd) EVENS/NUTTER LEXICAL RELATION HIERARCHY (i+,p-,nd) PUSTEJOVSKY'S QUALIA/EVENTS (i-,p-,nr,nd,c=Fass) FULTON'S SEMANTIC UNIFICATION METAMODEL (i-,p-,c=Fulton) RANDOM HOUSE WORD MENU CATEGORIES (i-,p-) PANGLOSS ONTOLOGY BASE (i-,p-,nr,nd) YALE'S ESPERANTO THESAURUS (i-,p-,nr,nd) DARPA/ROME PLANNING ONTOLOGY (i-,p-,nr,nd) CIMOSA BUSINESS ENTERPRISE ARCHITECTURE MODEL (I,p-) ICAM MANUFACTURING REFERENCE MODEL (i-,p-,nr,nd) IWI (GERMAN) MANUFACTURING REFERENCE MODEL (i-,p-,nr,nd) EDI (ELEC. DATA INTERCH.) BUSINESS COMM STANDARD(i-,p-,nr,nd) GUARINO CONCEPT/RELATION ONTOLOGY (i+,p-,nd) SKUCE ONTOLOGY (I,p-,c=Sarris) PETRIE ONTOLOGIES (i-,p-,nr,nd) TEPFENHART ONTOLOGY (i-,p-,nd) G.E. SEMANTIC HIERARCHY & LEXICON (RAU ET AL.) (i-,p-,nr,nd) PLINIUS CERAMICS ONTOLOGY (I,p-,c=Speel) MARS' KELVIN MEASUREMENT HIERARCHY (i-,p-,nr,nd,c=Speel) RANDELL & COHN'S SPATIOTEMPORAL LATTICES (I,p-) PDES/STEP PRODUCT DESCRIPTION STANDARDS (i+,p-,nr,nd) HARTLEY'S TIME AND SPACE WORLD (i+,p-,nd) ONTOLINGUA-KIF (i+,p-,nd,c=Gruber) GRUBER'S QUANTITIES/UNITS ONTOLOGY (i-,p-,nd,c=Gruber) GENERIC BIBLIOGRAPHY CONCEPT SYSTEM (i-,p-,nd,c=Gruber) DICK'S CASE-RELATION SYSTEM FOR LEGAL ANALYSIS (I,p-,c=Dick) HORN'S SUBJECT MATTER CATEGORIES (I,p-,c=Horn,nd) ARNOPOULOS' SYSTEM UNIFICATION MODEL (SUM) (I,p-,c=ISKO) TENENBAUM/EIT MANUFACTURING ONTOLOGY (i-,p-,nr,nd) INTELL. TEXT PROCESSING INC.SEMANTIC LEXICON (i-,p-,nr,nd) [Hundreds of special-area thesauri are listed in Int. Classification and Indexing Bibliography (ICIB) published by ISKO, the Int. Soc. for Knowledge Organization, Woogstr. 36a, D-6000 Frankfurt 50, Germany] SUGGESTED FORMAT: Brief description, Examples, Formalized?, Abstract hierarchy structure, Maximum depth, Necessary/sufficient?, Current authorities/enthusiasts, Machine-readable text?, Machine- usable structure?, Source, FTP site?, Implemented? References. I won't release the catalogue until at least half the entries are written up. ======================================================================== ______________________________________________________________________ >From chalmers@violet.berkeley.edu Sat Feb 12 00:07:21 1994 Received: from violet.Berkeley.EDU by odin.diku.dk with SMTP id AA06708 (5.65c8/IDA-1.4.4 for ); Sat, 12 Feb 1994 17:07:24 +0100 Received: from localhost by violet.berkeley.edu (8.6.4/1.33r) id IAA21770; Sat, 12 Feb 1994 08:07:21 -0800 Date: Sat, 12 Feb 1994 08:07:21 -0800 From: chalmers@violet.berkeley.edu (John H. Chalmers Jr.) Message-Id: <199402121607.IAA21770@violet.berkeley.edu> To: conlang@diku.dk Subject: Laadan Steve: Thanks for posting the La'adan class material. If you don't have the latest grammar, you might find the following of interest. I just received an answer from Suzette Haden Elgin re La'adan. The La'adan grammar and dictionary are available for $10.00 + $2.00 shipping and handling. For $6.00, there is a 60 minute cassette tape to accompany the grammar. She is the speaker. She has also made a VHS 60 minute videotape about La'adan and her books. It is $22.00 + $3.00 s/h. Order from the Ozark Center for Language Studies, PO Box 1137, Huntsville, AR 72740-1137 USA. (501) 559-2273. SHE says that the La'adan Network has not communicated with her for 2 years and she does not know its current status. Apparently all the money from the La'adan grammar goes SF3, defined as the Society for the Furtherance of and Study of Fantasy and Science Fiction, Inc. Box 1624, Madison, WI 53701-1624. From the xeroxes enclosed, "A First Dictionary and Grammar of La'adan 2nd edition appears to be a spiral bound book of 157+ pages. This format is quite different from the first book that was available. SHE also mentions that there are two La'adan Network Bulletins and that she will make them available if she can get permission. It appears that the slow pace of La'adan and poor communications has disappointed SHE since she contrasts her struggling women's project to the success of the "hypermasculine combat-focussed Klingon language." Now that all three of the Native Tongue books are back in print, there may be more interest in the language. Having the first two go out of print cannot have helped matters. I've uploaded this message from a local gateway in the San Diego area. I have not cancelled or moved my main address from Berkeley, reply to the List or to chalmers@violet.berkeley.edu -- John ______________________________________________________________________ >From ram@eskimo.com Sat Feb 12 19:08:44 1994 Date: Sun, 13 Feb 1994 03:08:44 -0800 From: ram@eskimo.com (Rick Morneau) Message-Id: <199402131108.AA12632@eskimo.com> To: conlang@diku.dk, ram@eskimo.com Subject: Re: Intro/AI/compound-nouns/concept-systems Fritz Lehmann writes: > > Because I'm so new I don't even know what language they are > talking about -- presumably one being devised by Rick Morneau. > (It sounds very interesting and I'd like to know where I could see > a description.) > There is no language (yet). I'm just studying how languages work, organizing what I'm learning, and trying to put it all in writing. If I ever DO design a language, my goals will be ease-of-learning (regardless of the background of the student), genetic neutrality, computer tractability, and linguistic realism, simplicity and naturalness. My ultimate goal is and always has been to develop a source/target-independent interlingua for use in machine translation. Regards, Rick *=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=* =* Rick Morneau ram@eskimo.com "All kings is rapscallions" *= *= Denizen of Idaho, USA --Mark Twain =* =*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*= ______________________________________________________________________ >From lojbab@access.digex.net Mon Feb 14 05:47:42 1994 From: Logical Language Group Message-Id: <199402141547.AA05701@access2.digex.net> Subject: Re: Intro/AI/compound-nouns/concept-systems To: conlang@diku.dk Date: Mon, 14 Feb 1994 10:47:42 -0500 (EST) Fritz Lehmann puts in his list: > LOGLAN/LOJBAN SEMANTIC PRIMITIVE WORD ROOTS (I,p-) I think some comment, or even protest, is needed here. Despite early claims by James Cooke Brown, it has long been clear (even to him) that the Loglan and Lojban gismu lists aren't "semantic primitives" in any sense. Rather, they are an attempt to blanket semantic space with a sufficient number of words, balancing various priorities such as terseness, primitiveness, and ease of use in constructing complexes (>tanru< in Lojban). To take an obvious example: Lojban has words for "female", "parent", and "mother". If "female" and "parent" are taken as semantic primitives, then "mother" cannot be semantically primitive: but in Lojban all three words are considered roots. (Furthermore, the words >rirni< 'parent' and >mamta< 'mother' refer to psychological-social parental relationships: there are separate forms for biological relations. Within the Lojban community (I don't speak for the Loglan Institute), the primary emphasis has been for some time the usefulness of a new root in the construction of >tanru<, and a vocal minority has proposed the deletion of various existing roots on the grounds that their usefulness for that purpose is minimal. (These efforts have been resisted to date.) -- John Cowan sharing account for now e'osai ko sarji la lojban. ______________________________________________________________________ >From shoulson@ctr.columbia.edu Mon Feb 14 06:00:53 1994 From: shoulson@ctr.columbia.edu (Mark E. Shoulson) Received: from localhost (shoulson@localhost) by startide.ctr.columbia.edu (8.6.5/8.6.4.788743) id LAA04391; Mon, 14 Feb 1994 11:00:53 -0500 Date: Mon, 14 Feb 1994 11:00:53 -0500 Message-Id: <199402141600.LAA04391@startide.ctr.columbia.edu> To: conlang@diku.dk Subject: Laadan >Date: Sat, 12 Feb 1994 17:21:32 +0100 >From: chalmers@violet.berkeley.edu (John H. Chalmers Jr.) >She has also made a VHS 60 minute videotape about La'adan and >her books. It is $22.00 + $3.00 s/h. Order from the >Ozark Center for Language Studies, PO Box 1137, Huntsville, AR >72740-1137 USA. (501) 559-2273. Tapes? Videos? Cool!! I must write to them. > Apparently all the money from the La'adan grammar goes SF3, >defined as the Society for the Furtherance of and Study of Fantasy >and Science Fiction, Inc. Box 1624, Madison, WI 53701-1624. From the >xeroxes enclosed, "A First Dictionary and Grammar of La'adan 2nd >edition appears to be a spiral bound book of 157+ pages. This format >is quite different from the first book that was available. I'll check which one I have; I recall it's also a spiral-bound puppey. > It appears that the slow pace of La'adan and poor communications >has disappointed SHE since she contrasts her struggling women's project >to the success of the "hypermasculine combat-focussed Klingon language." Oh, man, does that bug me. Klingon's success probably has little to nothing to do with the sexual equipment of its inventor; I suspect the fact that Star Trek has a *vast* fan network in place already and enormous name-recognition among non-aficcianadoes *might* have had something to do with the press and attention it gets, eh? The slightest bit larger than SHE's fan club, I fear. I'm struggling hard to retain the respect I have for SHE; she's nobody to sneeze at, and so far I'm winning. But this "new sexism" as I'm coming to call it really gets up my nose. >-- John ~mark ______________________________________________________________________ >From lojbab@access.digex.net Mon Feb 14 06:21:18 1994 From: Logical Language Group Message-Id: <199402141621.AA07028@access2.digex.net> Subject: Re: Intro/AI/compound-nouns/concept-systems To: conlang@diku.dk Date: Mon, 14 Feb 1994 11:21:18 -0500 (EST) Cc: fritz@rodin.wustl.edu, lojbab@access.digex.net (Logical Language Group) Fritz Lehmann writes: > I too am bothered by noun compounds in which the > relation between the primitive noun elements is "assumed". A > "mnemonic" composition isn't good enough, in my books. A > combination of "alligator" and "shoes" should differ from the one > of "horse" and "shoes" -- and "olive"+"oil" should differ from > "baby"+"oil". What's missing is usually the relevant case-like- > relation (like USED-FOR or PART-OF or AGENT-OF). It seems to me > that there are few enough of these basic relations that they could > be very compactly encoded (say with vowel-dipthongs). Your theory falls apart in the face of my facts :-) Ivan Derzhanski, Lojbanist and semanticist, has identified (using natural languages) some 34 such "case-like relations", with no claim of exhaustiveness. His languages (all of AN type) include: Abazin, Chinese, English, Ewe, Finnish, Georgian, Guarani, Hopi, Hungarian, Imbabura Quechua, Karaitic, Kazakh, Korean, Mongolian, Qabardian, Quechua, Russian, Sanskrit, Swedish, Turkish, Udmurt. Using your style, his "case-like relations" are roughly: OBJECT-OF-ACTION, TYPE-OF-SET-ELEMENTS, SET-OF-ELEMENTS-OF-TYPE, COMPONENT-OF, SPECIFIED-BY-DETAIL, SPECIES-OF, POSSESSED-BY, INHABITED-BY, CAUSE-OF, EFFECT-OF, INSTRUMENT-WITH-PURPOSE, OBJECT-OF-INSTRUMENT-OF-PURPOSE, PRODUCT-OF, SOURCE-OF, MADE-FROM, TYPICALLY-MEASURED-BY, ANALOGOUS-TO, ANALOGOUS-PART-OF, PRODUCER-OF, WITH-PROPERTIES-OF, RESEMBLING, CHARACTERISTICALLY-LOCATED-AT, SOLD-AT, APPLIED-TO, USED-AS-IMPLEMENT-DURING, PROTECTING-AGAINST, CHARACTERISTICALLY-CONTAINING, CHARACTERISTICALLY-CO-OCCURRING-WITH, SUPPLYING-ENERGY-TO, SPECIFYING-TEMPORAL-FRAME-OF, AND, OR, AND-AS-ALSO-TYPIFIED-BY, AND-ALSO-IS-IMPORTANT-PART-OF. Here are some examples corresponding to the relations above respectively (English examples have been used where possible; otherwise, see the Notes): pencil sharpener, row house, cell block, chicken feather, pendulum clock, pine tree, lion['s] mane, family land, tear gas, water mark, lamp shade, pepper stone (1), bear meat, coal mine, stone lion, land piece (2), space ship, tooth root, silk worm, soldier ant, cherry bomb, field mouse, book bar (3), tooth paste, Ping-Pong ball, rain hat, milk bottle, morning fog, electric lamp, milk tooth, home town, day night (4), worm beetle (5), land air (6). Notes: (1) Sanskrit, 'stone for grinding pepper' (2) Turkish, 'piece of land' (3) Chinese, 'bookstore/library' (4) Sanskrit, '24-hour day' (5) Mongolian, 'insect' (6) Finnish, 'world' > If oil is > BOR, OLIVE is BAK, and baby is BIK, I think BORaiBAK and BORiuBIK > are not much more burdensome on the speaker than BORBAK and > BORBIK, but vastly better if "ai" is "DERIVED-FROM" and "iu" is > something like "USED-FOR". I think the above list clearly demonstrates that the possible relationships between compounds approach, if not equal, in complexity the possibilities to be compounded. So we need to use our general list of roots to indicate the types of compounds -- which clearly leads to an infinite regress, as there are now a large number of second-order ways in which the linkage particles apply to the roots! And so on, and so on, >in saecula saeculorum<. Lojban cuts this process off at the start by simply saying that the relationships between compounds are not defined, and if one wishes to be more precise, one must be more explicit. lo cimni ka satci cu se jdima lo cimni ni valsi The price of infinite precision is infinite verbosity. > Case-relations are not just separate > entities -- they occur in a hierarchy of such relations. (See > A.F. Parker-Rhodes "Inferential Semantics", Harvester/Humanities > Pres, 1978, Chapter XI.) Some case-relationss are more general > than (i.e. they subsume) other more specific ones, so the case- > mark in compound nouns ned not refer to a too-specific case- > relation. But by saying this, you kick the ball through my goal-posts! There are compounds, as Ivan's treatment shows, which are discriminated from each other by just such precision. So either you admit polysemy in compounds, or you are forced into more and more "specific" case-relation markers to express increasing subtleties of concept. Disclaimer: Ivan is not responsible for the (mis)use I have made of his data, collected for an unpublished paper and used by permission. -- John Cowan sharing account for now e'osai ko sarji la lojban. ______________________________________________________________________ >From ucleaar@ucl.ac.uk Mon Feb 14 19:31:42 1994 From: ucleaar Message-Id: <116751.9402141931@link-1.ts.bcc.ac.uk> To: conlang@diku.dk Subject: Re: Laadan Date: Mon, 14 Feb 94 19:31:42 +0000 > > It appears that the slow pace of La'adan and poor communications > >has disappointed SHE since she contrasts her struggling women's project > >to the success of the "hypermasculine combat-focussed Klingon language." > > Oh, man, does that bug me. Klingon's success probably has little to > nothing to do with the sexual equipment of its inventor; I suspect the fact > that Star Trek has a *vast* fan network in place already and enormous > name-recognition among non-aficcianadoes *might* have had something to do > with the press and attention it gets, eh? The slightest bit larger than > SHE's fan club, I fear. I'm struggling hard to retain the respect I have > for SHE; she's nobody to sneeze at, and so far I'm winning. But this "new > sexism" as I'm coming to call it really gets up my nose. Surely part of the appeal of Klingon is its iconic & ironic brutalism. Just as stories about peace & love tend to be less exciting than stories about struggle & strife, so perhaps the discordance & rudeness of Klingon leads to it being preferred over a more civil language. John doesn't quote much context from Elgin's letter, but she seems to be making a point feminists often, with reason, make, e.g. "Why do men prefer watching films about war rather than about babies" etc. etc. Fair enough, no? ---- And ______________________________________________________________________ >From chalmers@violet.berkeley.edu Mon Feb 14 12:48:56 1994 Date: Mon, 14 Feb 1994 20:48:56 -0800 From: chalmers@violet.berkeley.edu (John H. Chalmers Jr.) Message-Id: <199402150448.UAA11171@violet.berkeley.edu> To: conlang@diku.dk Subject: L'aadan If anything SHE's comments are more bitter in context. Let me quote the entire paragraph: " I do know there are small groups working with L'aadan in this country and in Canada; I hear about them from time to time. I know that the grammar is used in a number of university courses, for various purposes. I still sell small quantities of L'aadan products, and get an occasional query. However, there is nothing like the swell of support that was available for the hypermasculine combat-focussed Klingon language, from Star Trek -- no big commercial publisher for the book and the tape, no media push, no best seller status, no summer camp, no university affiliation with a scholarly journal. I'm sorry about that, but there you are; it's not the first time (she says understatedly) that a women's project has struggled just to survive while a comparable men's project has thrived. I tell you this not to complain -- I have always said that L'aadan had to stand or fall on its own, like any other language-- but you deserve the most honest answer I'm able to provide." So, I would urge Stephen and others to let her know that somebody out there is still interested in her creation, even if they are men. Unfortunately, I don't think Native Tongue III is going to bring many new recruits into the fold as there is not enough about L'aadan in it. To address And's point, I think it is not just the attraction and excitement of violence, but rather that in Western culture, more men than women are interested in intellectual matters. I speak from some experience; I've seldom, if ever, found a woman very interested in intellectual fields other than her own profession. I've given up trying to get dates for contemporary music concerts, for example, though I know a number of women composers and performers of such music (all happily married alas). I don't know the reason, but it must be due to early socialization and schooling. --John ______________________________________________________________________ >From jrk@sys.uea.ac.uk Tue Feb 15 09:44:35 1994 Date: Tue, 15 Feb 94 09:44:35 GMT Message-Id: <17295.9402150944@s5.sys.uea.ac.uk> To: conlang@diku.dk From: jrk@sys.uea.ac.uk (Richard Kennaway) Subject: Re: Intro/AI/compound-nouns/concept-systems John Cowan writes: >I think some comment, or even protest, is needed here. Despite early >claims by James Cooke Brown, it has long been clear (even to him) that >the Loglan and Lojban gismu lists aren't "semantic primitives" in any sense. >Rather, they are an attempt to blanket semantic space with a sufficient >number of words, balancing various priorities such as terseness, primitiveness, >and ease of use in constructing complexes (>tanru< in Lojban). I think some comment, or even protest, is needed here. I don't speak for the Loglan Institute, but the sly put-down of JCB above is quite unwarranted. The "early claims" must be early indeed, for I have never seen them in my contact with Loglan. It is quite clear from, e.g. the 1975 edition of Loglan 1, that the "primitive" predicates are so called only because they are syntactically primitive, and that their choice was driven primarily by the need "to blanket semantic space with a sufficient number of words", primitive and complex. -- ____ Richard Kennaway __\_ / School of Information Systems jrk@sys.uea.ac.uk \ X/ University of East Anglia \/ Norwich NR4 7TJ, U.K. ______________________________________________________________________ >From lojbab@access.digex.net Tue Feb 15 05:39:59 1994 From: Logical Language Group Message-Id: <199402151539.AA20127@access2.digex.net> Subject: Re: Intro/AI/compound-nouns/concept-systems To: conlang@diku.dk Date: Tue, 15 Feb 1994 10:39:59 -0500 (EST) Richard Kennaway writes: > I think some comment, or even protest, is needed here. I don't speak for > the Loglan Institute, but the sly put-down of JCB above is quite > unwarranted. No put-down of JCB was intended. However, he is (to say the least) a man well-known for not changing his mind easily: although Loglan '60 is quite different from Loglan '91, a great many things persist absolutely unchanged despite well-established difficulties discovered by the rest of the Loglan community (and I mean things found before the Loglan/Lojban split). As the joke has it, "I am firm, you are stubborn, he is a pigheaded fool." > The "early claims" must be early indeed, for I have never > seen them in my contact with Loglan. It is quite clear from, e.g. the 1975 > edition of Loglan 1, that the "primitive" predicates are so called only > because they are syntactically primitive, and that their choice was driven > primarily by the need "to blanket semantic space with a sufficient number > of words", primitive and complex. Correct. I did say "has long been clear". However, in >Loglan 2< (1970) we are told that the aboriginal source for the prim/gismu list was "all those concepts which had been found to be universal in human languages. There are some 200 of these: words like 'man', 'woman', 'eat', 'vomit', and other rather earthy words." The impersonal passive doesn't tell us where these words came from, but (despite the large figure of 200) they seem to have been "semantic primitives" in JCB's mind: he also calls them "anthropologically universal" and "universal in human languages". (L2 ch. 3, repr. TL1:5/305-06). Of course, although L2 was first published in 1970, and TL1:5 in 1977, JCB is speaking here of work done before the Scientific American publication of 1960, so "early indeed" is a suitable characterization. -- John Cowan sharing account for now e'osai ko sarji la lojban. ______________________________________________________________________ >From lojbab@access.digex.net Tue Feb 15 13:03:24 1994 Date: Tue, 15 Feb 1994 18:03:24 -0500 From: Logical Language Group Message-Id: <199402152303.AA20695@access3.digex.net> To: conlang@diku.dk Subject: Re: Intro/AI/compound-nouns/concept-systems Cc: lojbab@access.digex.net X-Charset: LATIN1 X-Char-Esc: 29 >I think some comment, or even protest, is needed here. I don't speak >for the Loglan Institute, but the sly put-down of JCB above is quite >unwarranted. I'm sorry this came across as a put-down of JCB. Maybe those of us at LLG are suspect in this area, but I am fairly sure John did not intend a put-down. It was a simple statement of fact. JCB has made such claims, both in print, and to me personally, in debating issues regarding what words should be in the prim/gismu list. And it is equally clear that his thinking has evolved, though contrary to John, I think that JCB still DOES consider semantic primitiveness a quality of his set of 'composite primitives'. I actually think the evolution in JCB's thinking has gone the other way - he originally was more open to the idea the the primitives were more syntactic than semantic, but opposition, including ours in the community that eventually became LLG (pc and myself being major advocates against interpreting primitives as semantic primitives), has hardened his position and closed his mind. Recent Documentation: Loglan I 4th edition (1989), p411 "The composite primitives of Loglan are the semantically universal predicates of human experience." I can't imagine a much clearer statement than that one. He then goes on to state that adding new C-prims is rare indeed, using the late-added distinction in Loglan between "setci" (set) and "klesi" (class) as an example: "The distinction between sets and classes is semantically fundamental." There is thus no doubt that JCB considers his composite-primitive set to be a set of semantic primitives. The *only* contention that counters this general observation is that TLI Loglan no longer has clear boundaries between the composite-primitives - the ones made from 8 languages, and the borrowings and invented words thta have the same morphology, but for which JCB would make no claim of semantic primitiveness. As for more complete discussion that you, Richard may have available, I will cite a longish essay on the subject in The Loglanist 3/3, pg 219-21. I will give only some extracted quotes - JCB's thought processes are made extremely clear in that essay. I will try to outline his reasoning as I go. He was responding to a proposal to add a lot of primitives for zoological hierarchies, which proposal also argued that some primitives should be compounds based on those primitives; e.g. man = male-adult-human; mother = female-parent. JCB was vociferously opposed. He accepts the usefilness of the zoological terminology "not as a replacement for our vocabulary of 'people' primitives, but as a scientific adjunct to it, perhaps with its own set of primtiives". "... For there reappears in RLW's argument ... a pernicious idea ... one thta apparently dies very hard among loglanists: namely, that 'primitiveness' is whatever produces definitional efficiency, i.e. shared parsimony. There are many criteria of 'primitiveness and this is one of them, the logician's. Another, perhaps more appropriate to an empirical science, is 'whatever carves reality at the joints' ... [comment that this phrase is a quote from some unrecalled academic source]. Of course, to carve reality at the joints one must first know where the joints lie, that is, do science ... And it should, therefore be our guiding principle in incorporating their [the sciences'] vocabularies into Loglan. ..." It seems clear that these two categories of primitiveness are those associated with 'dividing up of semantic space', and JCB is saying they are important but are NOT the proper basis for the language. "But there is a third notion of primitiveness that is even more directly related to our concerns as language-builders, and that is what I've called 'ethological primitiveness'..." He then goes on to argue that by logical or scientific primitiveness, "girl" is not primitive, but is "female-child". But this kind of breaking into "components ... are not built into the neurological equipment of the species ... On the other hand, *whole* boys, girls, men, women, fathers, mothers, snakes and horses *are* built into that neurological substrate of human perception...or so I have hypothesized, else it would not be the case that nearly all human tongues, despite their other diversity, hail objects like these, and a few hundred others, with etymologically simple and mutually unrelated terms." He then goes on to compare this hypothesis of his with the fundamentalness of the 'deep grammar' hypothesis of Chomsky, both "supported by the weight of the contemporary evolutionary evidence" - like Chomsky did, he argues not from direct research on the subject, but "there hardly seems to be any other way in which these massive uniformities could have developed". [lojbab: I didn't buy this argument then, and still don't. There are plausible explanations that do not require any particular set of primitive concepts to be built into our genes. But JCB clearly believes that there ARE universal semantic primitives, and designed his language on that basis.] The major reason why people may come to believe that JCB feels otherwise, is the way he came to select his set of primitives, which is to some extent fundamentally inconsistent with his theories and pronouncements. I won't quote the wordy and not-very-to-the-point passages from Loglan II, which were reprinted in TL volume 1, on the selection of the set of primitives, but it involved several steps. JCB first used sets of semantic primitives universal to 'all' languages based on research that had been done in that area (he has never given citations for such research). He then used Ogden's Basic English as an additional source of primitives - perhaps not realizing that Ogden WAS looking for syntactic rather than semantic primitives. But then he justifies using Basic English because of Ogden's demonstrated success in "rendering ... almost any English text", which sounds like a semantic-space coverage argument to me. Thirdly, he used Eaton's semantic frequency list, making sure that the first 600 or so words were prims, even if they were not in the list for other reasons. Then, for the words up to frequency 1000, he required that they be expressible as either primitive or at most 2-parts, and from 1000-5000 rank, at most 3 parts. He then added some prims because they were "metaphorically productive", in making a lot of other compounds shorter, even if they were not particularly primitive by any other rationale. Now in the 30-odd years since then, he has added perhaps 100-200 words to his primitive list. Many of these were not composite primitives, made from 8 languages, but rather borrowings and scientific words. The largest sets added were 1) culture words 2) miscellaneous concrete terms for everday use like foods, games (football and billiards) added as "International prims" in 1972-5, 3) a group of words of the 'ethnological primitive' variety - specifically words pertaining to human bodily function that are commonly taboo and hence were not listed in research works earlier in this century (this in response to the criticisms of Arnold Zwicky in his review of Loglan I in 1969 in the journal 'Language'), 4) metric words, and 5) a bunch of words added as a result of Faith Rich's completing the analysis of Eaton's semantic frequency work - these are a combination of words of the 'semantically productive' variety mentioned above, and words that Faith couldn't come up with good metaphors for. In reinventing the language as Lojban, we started with JCB's list as a given, but then used the metaphor of 'covering semantic space' as a rationale for our own extensions to the list. As I recall, that metaphor first was presented by a local-area Loglanist who has been a long time lurker following the project. The idea of blanketing semantic space with small contiguous 'tiles' representing words was his personal concept of how Loglan semantics, with unitary word meanings, should be. We ttok up his metaphor and talked this up a lot in the early years of redeveloping Lojban, but JCB never recognized, adopted, or otherwise used words like 'covering semantic space' in any of his writings that I can recall, and that usage at the first DC LogFest was the first time I ever heard that phrase used. (I've since seen it used a lot here on conlang, of course). lojbab ---- lojbab lojbab@access.digex.net Bob LeChevalier, President, The Logical Language Group, Inc. 2904 Beau Lane, Fairfax VA 22031-1303 USA 703-385-0273 ______________________________________________________________________ Date: Wed, 16 Feb 94 10:41:04 CST From: fritz@rodin.wustl.edu (Fritz Lehmann) Message-Id: <9402161641.AA09516@rodin.wustl.edu> To: conlang@diku.dk Subject: combining nouns via case; primitives X-Charset: LATIN1 X-Char-Esc: 29 Dear ConLangers, I got two criticisms replying to my introductory note. John Cowan said: >Ivan Derzhanski, Lojbanist and semanticist, has identified (using natural >languages) some 34 such "case-like relations", with no claim of exhaustiveness. >His languages (all of AN type) include: [languages omitted] >Using your style, his "case-like relations" are roughly: >OBJECT-OF-ACTION, TYPE-OF-SET-ELEMENTS, SET-OF-ELEMENTS-OF-TYPE, COMPONENT-OF, >SPECIFIED-BY-DETAIL, SPECIES-OF, POSSESSED-BY, INHABITED-BY, CAUSE-OF, EFFECT-OF, >INSTRUMENT-WITH-PURPOSE, OBJECT-OF-INSTRUMENT-OF-PURPOSE, PRODUCT-OF, SOURCE-OF, >MADE-FROM, TYPICALLY-MEASURED-BY, ANALOGOUS-TO, ANALOGOUS-PART-OF, PRODUCER-OF, >WITH-PROPERTIES-OF, RESEMBLING, CHARACTERISTICALLY-LOCATED-AT, SOLD-AT, >APPLIED-TO, USED-AS-IMPLEMENT-DURING, PROTECTING-AGAINST, >CHARACTERISTICALLY-CONTAINING, CHARACTERISTICALLY-CO-OCCURRING-WITH, >SUPPLYING-ENERGY-TO, SPECIFYING-TEMPORAL-FRAME-OF, AND, OR, >AND-AS-ALSO-TYPIFIED-BY, AND-ALSO-IS-IMPORTANT-PART-OF. [examples omitted] This is a very interesting list and I would like to get Ivan Derzhanski's original research on this. Can anyone send this to me? Or his address? I'll probably add him to the master list of concept- systems. I believe that careful development of deep-case-like relations should be a first priority of a language creator. My list of concept- systems included several case systems. These include Fillmore's deep cases (and derivatives), the list of about 28 cases of Karen Sparck Jones and Branimir Boguraev, Somers' "case grid", the lattice of cases in Parker-Rhodes' Inferential Semantics, Martha Evens & Terry Nutter's hierarchy of relations, and Judy Dick's dissertation on case-relations in legal reasoning. No doubt there are others. Is there in existence a bibliography, or comparative work discussing semantic case systems? >> If oil is >> BOR, OLIVE is BAK, and baby is BIK, I think BORaiBAK and BORiuBIK >> are not much more burdensome on the speaker than BORBAK and >> BORBIK, but vastly better if "ai" is "DERIVED-FROM" and "iu" is >> something like "USED-FOR". > I think the above list clearly demonstrates that the possible relationships >between compounds approach, if not equal, in complexity the possibilities >to be compounded. So we need to use our general list of roots to indicate >the types of compounds -- which clearly leads to an infinite regress, as >there are now a large number of second-order ways in which the linkage >particles apply to the roots! And so on, and so on, >in saecula saeculorum<. I mentioned two-letter combinations (maybe just vowels). This would cover the necessary combinations. 34 or 28 is not too many. Seven vowels might do it. Also, rare relations might have longer combinations so there could be more. >> Case-relations are not just separate >> entities -- they occur in a hierarchy of such relations. (See >> A.F. Parker-Rhodes "Inferential Semantics", Harvester/Humanities >> Press, 1978, Chapter XI.) Some case-relations are more general >> than (i.e. they subsume) other more specific ones, so the case- >> mark in compound nouns need not refer to a too-specific case- >> relation. >But by saying this, you kick the ball through my goal-posts! There are >compounds, as Ivan's treatment shows, which are discriminated from each >other by just such precision. So either you admit polysemy in compounds, >or you are forced into more and more "specific" case-relation markers >to express increasing subtleties of concept. No. You can legitimately choose any level of generality within the poset of all relations -- you're not forced to admit anything. Loglan/Lojban chose the very top of the poset: juxtapositions mean SOME-RELATION which is the least informative choice possible. A diligent scholar could discriminate more than 50 case-like relations (although they will cease to be case-like at about that point e.g. relations like ALONG-THE-FORMER-PATH-OF-THE SHADOW-OF will begin to occur). I recommend using a more specific level than SOME-RELATION so as to eliminate most of the guesswork in recognizing the meaning of a compound. Generality is not "polysemy"; outside of logic and mathematics every word is approximative in that a more specific description is possible. Learning 10-40 relation-morphemes would be a burden to start with, but thereafter the payoff would be big. The other criticism was of my including LOGLAN/LOJBAN SEMANTIC PRIMITIVE WORD ROOTS on my concept-systems list. The subsequent email discussion tells me that no change is needed. When I say "semantic primitive" I don't necessarily mean ultimate philosophical primitives. I happen to share Leibniz's desire for ultimate primitives, but although maybe half of the concept systems have "primitives" from which all other concepts are defined, very few (maybe Leibniz, Burger, Wierzbicka, Schank, Skuce and Tepfenhart) aim for ultimate primitives. Almost all concept- systems with definitions have the notion of "more primitive" though. The notion of "covering the semantic space" involves a set of primitive concepts _sufficient_ to say anything. This is rarely claimed to be a _necessary_minimal_ set. Yours truly, Fritz Lehmann 4282 Sandburg, Irvine, CA 92715 714-733-0566 fritz@rodin.wustl.edu ==================================================================== ______________________________________________________________________ >From doug@netcom.com Wed Feb 16 01:24:35 1994 Message-Id: <199402161724.JAA13601@mail.netcom.com> From: doug@netcom.com (Doug Merritt) Date: Wed, 16 Feb 1994 09:24:35 PST "Re: Intro/AI/compound-nouns/concept-systems" (Feb 16, 12:19am) To: conlang@diku.dk Subject: Re: Intro/AI/compound-nouns/concept-systems X-Charset: LATIN1 X-Char-Esc: 29 Lojbab quoted: >Loglan I 4th edition (1989), p411 >"The composite primitives of Loglan are the semantically universal >predicates of human experience." That's an unfortunate attitude. If there are semantic universals, we have an exceedingly ill-defined idea of what they are with the current state of the art. One of the best studies on the subject I ever saw was the book from about 10 to 15 years ago "Cross-Cultural Universals of Affective Meaning", in which both empirical studies and the following statistical analysis were *very* carefully designed. But even this gave fairly slippery results, certainly very little to support a claim like JCB appears to be making. Eventually further studies could be done using e.g. newer NMR brain imaging, and there's some interesting work being done in neurophysiology and such that indirectly helps out with the topic. But we're still quite some ways away from having "semantically universal predicates of human experience." >shared parsimony. There are many criteria of 'primitiveness and this is >one of them, the logician's. Another, perhaps more appropriate to an >empirical science, is 'whatever carves reality at the joints' ... And in *this* sense there will never be a universal set of primitives, because in addition to whatever is hardwired by biology, there is a further set that is based on both culture and experience, and those things are primitives only in a subset of the human race. Doug ______________________________________________________________________ >From lojbab@access.digex.net Wed Feb 16 08:17:42 1994 From: Logical Language Group Message-Id: <199402161817.AA07372@access2.digex.net> Subject: Re: combining nouns via case; primitives To: conlang@diku.dk Date: Wed, 16 Feb 1994 13:17:42 -0500 (EST) Cc: lojbab@access.digex.net (Logical Language Group), iad@cogsci.ed.ac.uk X-Charset: LATIN1 X-Char-Esc: 29 Fritz Lehmann writes: > This is a very interesting list and I would like to get Ivan > Derzhanski's original research on this. Can anyone send this to me? Or > his address? I'll probably add him to the master list of concept- > systems. Ivan is . I can't give you his paper, because it is unpublished and I have no right to distribute it. (But I did send him a copy of my reply, and will cc this message to him as well. Hi, Ivan.) > I believe that careful development of deep-case-like relations > should be a first priority of a language creator. I think so too, provided it can be done. My intention in displaying Ivan's list was to claim that there is no algorithm for developing such a list; that it is as arbitrary as the list of roots to be compounded. > I mentioned two-letter combinations (maybe just vowels). This > would cover the necessary combinations. 34 or 28 is not too many. > Seven vowels might do it. Also, rare relations might have longer > combinations so there could be more. But Ivan explicitly disclaimed completeness. Another example: in Li & Thompson's >Mandarin Chinese: A Functional Reference Grammar (1981,1989: Univ. of California Press, ISBN 0-520-04286-7 cloth, 0-520-06610-3 pbk), Section 3.2.2 consists of an substantial list of Chinese nominal compounds sorted by relations. (Looking this over, I see that some of the relationships I attributed to Ivan are really L & T's. Apologies to all involved.) I think the summary of this section is worth quoting: The above list of 21 types of nominal compounds by no means constitutes an exhaustive categorization; one can still think of nominal compounds that are not accounted for in the above listing. The important thing to note, though, is that the compounding process of linking noun and noun together ... is a productive and creative one. As we said before, the only constraint is a pragmatic one, and that is that the context must be appropriate for naming a certain object.... > You can legitimately choose any level of generality > within the poset of all relations -- you're not forced to admit > anything. Loglan/Lojban chose the very top of the poset: > juxtapositions mean SOME-RELATION which is the least informative > choice possible. A diligent scholar could discriminate more than > 50 case-like relations (although they will cease to be case-like at > about that point e.g. relations like ALONG-THE-FORMER-PATH-OF-THE > SHADOW-OF will begin to occur). So you have a definite criterion for "case-like-ness"? I don't believe there is such a criterion, but I would love to see someone exhibit one. Most people who believe in fixed case lists seem to blow off that question with Justice Stewart's comment on obscenity: "I know it when I see it." In other words, the definition belongs to the informal level of the (case-)linguistic subculture. (See Hall.) Can this definition be "surfaced" into a formal- or technical-level one? > I recommend using a more specific > level than SOME-RELATION so as to eliminate most of the guesswork > in recognizing the meaning of a compound. See comment below. > Generality is not > "polysemy"; outside of logic and mathematics every word is > approximative in that a more specific description is possible. I agree with this statement (at last, he thinks :-)). However, I suspect that any short list of cases will contain ones that are themselves polysemous, and therefore lead to compounds polysemous by any standard. Without a list to talk about, however, this suspicion remains empty. > Learning 10-40 relation-morphemes would be a burden to start with, > but thereafter the payoff would be big. Lojban does have related machinery, although not in the specification of compounds. Instead, we have an extensible set of case tags which may be attached to arguments. As is well known (:-)), every Lojban predicate has a definite set of arguments whose meaningfulness is asserted by every use of the predicate: predicating >klama< 'come,go' of something entails the existence of a goer, a destination, an origin, a route, and a means. In addition, however, there is an indefinite (not infinite) set of additional arguments which can be made explicit by attaching an argument-expression using a case tag. There are 60-odd premade case tags, plus 20-odd more that express spatio-temporal relationships. However, there is also syntactic machinery for making any argument-slot of any statable relationship (using the full grammar of relationships, not just individual words) into a case tag. (In Lojban jargon: every selbri can be made into a tag by prefixing "fi'o".) > The other criticism was of my including LOGLAN/LOJBAN SEMANTIC > PRIMITIVE WORD ROOTS on my concept-systems list. The subsequent > email discussion tells me that no change is needed. When I say > "semantic primitive" I don't necessarily mean ultimate > philosophical primitives. Okay, given your clarification [which is omitted], I suppose you are correct. The Loglan Project tends to be allergic to the phrase "semantic primitives": our primitives are meant to be syntactically primitive only, in the sense that all compounds whether open or closed are built up from them. > Almost all concept- > systems with definitions have the notion of "more primitive" > though. Insofar as the Lojban gismu list is a set of primitives at all, all members of it are considered equal in "degree of primitiveness", though not perhaps in pragmatic usefulness. > The notion of "covering the semantic space" involves a set > of primitive concepts _sufficient_ to say anything. This is rarely > claimed to be a _necessary_minimal_ set. Just so. Lojbanists' view of case theory is that, for our purposes, no sufficiently compelling version of it exists. The other wing of the Loglan Project has a fixed list of 13 case tags, and a supposed assignment of each standard argument-slot of each (syntactically) primitive predicate to one of the 13, but the quality of this work has been questioned by people on both sides of the divide. -- John Cowan sharing account for now e'osai ko sarji la lojban. ______________________________________________________________________ >From lojbab@access.digex.net Wed Feb 16 08:33:30 1994 From: Logical Language Group Message-Id: <199402161833.AA07823@access2.digex.net> Subject: Re: Intro/AI/compound-nouns/concept-systems To: conlang@diku.dk Date: Wed, 16 Feb 1994 13:33:30 -0500 (EST) X-Charset: LATIN1 X-Char-Esc: 29 James Cooke Brown wrote: > >There are many criteria of 'primitiveness and this is > >one of them, the logician's. Another, perhaps more appropriate to an > >empirical science, is 'whatever carves reality at the joints' ... Doug Merrit writes: > And in *this* sense there will never be a universal set of primitives, > because in addition to whatever is hardwired by biology, there is a further > set that is based on both culture and experience, and those things are > primitives only in a subset of the human race. Ironically, I think this phrase is Whorf's, and if so, JCB is using it in a non-Whorfian and perhaps anti-Whorfian sense. The "joints" of which Whorf spoke were culturally based, but JCB talks of them as if they were hardwired by The Nature Of Things; i.e. physics and such, not hominoid biology, still less hominoid culture! Sidenote: I will no longer talk of "human rights", but of "hominoid rights". Hominoid are they born to Hominoidal Mitochondrial Eve. -- John Cowan sharing account for now e'osai ko sarji la lojban. ______________________________________________________________________ >From hrick@world.std.com Wed Feb 16 14:51:14 1994 Date: Wed, 16 Feb 1994 19:51:14 -0500 From: hrick@world.std.com (Rick Harrison) Message-Id: <199402170051.AA24239@world.std.com> To: conlang@diku.dk Subject: in defense of natlang-style compounding X-Charset: LATIN1 X-Char-Esc: 29 While "it" might be interesting in an experimental conlang, with regard to international auxiliary languages I oppose explicitly marking the relationships between radicals in a compound word for three reasons: "it" is difficult, inefficient and unnecessary. (Through the rest of this message, the pronoun "it" will refer to "the practice of marking the relationships between radicals in a compound word.") 1. difficult Looking at Derzhanski's list of case-like relationships, I felt that some of them could be merged into fewer items, while others could be divided into finer-grained divisions. Other readers probably had the same thought but would come up with a different list from mine. I doubt that two or more people could ever reach complete agreement on a list of relationships, or that any single conlanger could sustain whole-hearted support for a given list for any length of time. One person might argue that the relationship in "chicken-feather" is `component of,' while someone else might (equally reasonably) say the relationship is `inhabited by' or `produced by.' "It" opens the door to endless quibbles and variations. 2. inefficient "It" requires the insertion of extra phonemes into a compound in order to indicate the relationship. Thus "it" lengthens words and decreases efficiency, while providing little or no benefit in return; akin to flexion of adjectives to indicate case, or flexion of verbs to indicate person and number. 3. unnecessary Natlangs that engage in compounding don't do "it."* In practice, people of normal and even subnormal intelligence have no trouble understanding the meaning of "snowman" or "graveyard" or "rowboat." Compounding is a powerful tool for expanding the usefulness of a limited stock of radicals. I see no reason to cripple this tool by embracing "it." *Rick Morneau posted some French phrases which he called compounds, and which he offered as evidence of a natlang doing "it." I opine that they are not such evidence, as a) they are set phrases and not compounds, and b) natlang prepositions are so idiomatic and polysemous that they could just as well be replaced by grunts or silent hyphens or unmarked juxtaposition. ______________________________________________________________________ >From fritz@rodin.wustl.edu Thu Feb 17 00:30:46 1994 Date: Thu, 17 Feb 94 06:30:46 CST From: fritz@rodin.wustl.edu (Fritz Lehmann) Message-Id: <9402171230.AA10483@rodin.wustl.edu> To: conlang@diku.dk Subject: RE: in defense of natlang-style compounding X-Charset: LATIN1 X-Char-Esc: 29 hrick@world.std.com (Rick Harrison) said: >While "it" might be interesting in an experimental conlang, with regard >to international auxiliary languages I oppose explicitly marking the >relationships between radicals in a compound word for three reasons: >"it" is difficult, inefficient and unnecessary. (Through the rest of >this message, the pronoun "it" will refer to "the practice of marking >the relationships between radicals in a compound word.") > 1. difficult >Looking at Derzhanski's list of case-like relationships, I felt that >some of them could be merged into fewer items, while others could be >divided into finer-grained divisions. Other readers probably had the >same thought but would come up with a different list from mine. I doubt >that two or more people could ever reach complete agreement on a list of >relationships, or that any single conlanger could sustain whole-hearted >support for a given list for any length of time. Yes, I specifically said that you can choose the level of generality. I suggest doing so. You have to do the same in choosing your basic vocabulary and in everything else. The fact that interested people may disagree on where the line should be drawn doesn't mean it shouldn't be done at all. >2. inefficient >"It" requires the insertion of extra phonemes into a compound in order >to indicate the relationship. Thus "it" lengthens words and decreases >efficiency, while providing little or no benefit in return; akin to >flexion of adjectives to indicate case, or flexion of verbs to indicate >person and number. All true, except the "no benefit". If the otherwise semantically opaque becomes clear, that's benefit. I'd be interested to know the real cost. >3. unnecessary >Natlangs that engage in compounding don't do "it."* In practice, people >of normal and even subnormal intelligence have no trouble understanding >the meaning of "snowman" or "graveyard" or "rowboat." Compounding is a >powerful tool for expanding the usefulness of a limited stock of radicals. >I see no reason to cripple this tool by embracing "it." >*Rick Morneau posted some French phrases which he called compounds, >and which he offered as evidence of a natlang doing "it." I opine that >they are not such evidence, as a) they are set phrases and not compounds, >and b) natlang prepositions are so idiomatic and polysemous that they >could just as well be replaced by grunts or silent hyphens or unmarked >juxtaposition. I feel that b) almost cinches my case. Maybe the very worst thing about learning a natural language is learning the idiosyncrasies of nonsensical case-relations. Pretty good non-native English-speakers have told me how hard it is to figure out the correct English preposition. They want to say "look to" but it's "look for". This is no merit. It is a strong argument for using a planned language instead of a natural one. You say "people have no trouble understanding the meaning." That's true for _natives_, steeped in the culture, but not for strangers. If I visit the jungle and hear about spirit bananas, square jerks, river lines, thatch endeavors, crocodile rings, earth mothers, mother earths, snow men, monkey beer, etc., I will be at a loss. And, although you don't aim to address the computer and artificial intelligence aspect, any processes of logical elimination "obvious" to some native people ("alligator shoes must mean something other than shoes they wear because alligators don't wear shoes") will be extremely hard for computers even when they have enormous built-in "ontologies" and factual knowledge. Yours truly, Fritz Lehmann 4282 Sandburg Way, Irvine, California 92715, U.S.A. Tel.: (714)-733-0566 Fax: (714)-733-0506 fritz@rodin.wustl.edu ==================================================================== ______________________________________________________________________ >From lojbab@access.digex.net Thu Feb 17 04:45:07 1994 Date: Thu, 17 Feb 1994 09:45:07 -0500 From: Logical Language Group Message-Id: <199402171445.AA03873@access1.digex.net> To: lojbab@access.digex.net Subject: Re: combining nouns via case; primitives Cc: conlang@diku.dk X-Charset: LATIN1 X-Char-Esc: 29 LL> Lojbanists' view of case theory is that, for our purposes, no sufficiently LL> compelling version of it exists. The other wing of the Loglan Project LL> has a fixed list of 13 case tags, and a supposed assignment of each LL> standard argument-slot of each (syntactically) primitive predicate to one LL> of the 13, but the quality of this work has been questioned by people on LL> both sides of the divide. LL> LL> -- LL> John Cowan sharing account for now LL> e'osai ko sarji la lojban. Amending that slightly. Our resident logician, John Parks-Clifford, specifi- cally researched case theory as it stood in 1988 or so when we were doing that par tof the redesign. His conclusion was that the leading theorists were of the opinion that, while there was a possibility for a deep structure theory of cases, that theory allowed for an effectively open set of cases. Again, somewhere I should have a biblio entry on his major source. lojbab lojbab@access.digex.net ______________________________________________________________________ >From lojbab@access.digex.net Thu Feb 17 04:39:21 1994 Date: Thu, 17 Feb 1994 09:39:21 -0500 From: Logical Language Group Message-Id: <199402171439.AA03634@access1.digex.net> To: conlang@diku.dk Subject: Re: combining nouns via case; primitives Cc: lojbab@access.digex.net X-Charset: LATIN1 X-Char-Esc: 29 Re: Fritz's list of concept systems: if you are interested in the literature on relations between compoiunds, of the sort that Ivan did, I believe I have a couple of files of bibliogrphic entries on the subject. It is a heavily researched field, and I'm sure several of the othe papers and books in the bibliography have variations on such lists as Ivan's. I can send it to you personally, or post it to all of conlang if there is sufficient interest. lojbab lojbab@access.digex.net ______________________________________________________________________ >From lojbab@access.digex.net Thu Feb 17 05:24:42 1994 From: Logical Language Group Message-Id: <199402171524.AA06863@access2.digex.net> Subject: Re: in defense of natlang-style compounding To: conlang@diku.dk Date: Thu, 17 Feb 1994 10:24:42 -0500 (EST) Cc: lojbab@access.digex.net (Logical Language Group) X-Charset: LATIN1 X-Char-Esc: 29 Rick Harrison writes: > In practice, people > of normal and even subnormal intelligence have no trouble understanding > the meaning of "snowman" or "graveyard" or "rowboat." While I agree with most of Rick's post, I must take issue with this sentence. Your theory, like Fritz Lehmann's, falls apart in the face of my facts. :-) While it is true that compounds like "mouse trap", "rain drop", and "milk tooth" are understood in many languages, there are other compounds that are given very different interpretations in different languages. For example, "tea cup" means the same in English and Chinese, but in Korean and Abazin it means "cup of tea"; it specifies a quantity, not a physical object. Similarly, in English and Hungarian, "glass eye" means an eye which is made from glass, but in Quechua the same compound signifies spectacles, a glass adjunct to a (flesh) eye. Remember that all these languages use adjective- noun order, unlike (say) French or Spanish. I think English-speakers would need a lot more than "normal [or] even subnormal intelligence" to figure out Mongolian's compound "worm beetle" = "insect", or Kazakh's "cup plate" = "crockery". These compounds are of the type in which the two words being compounded provide typical examples of a more inclusive class which is the meaning of the compound itself. I point, in addition, to the Lojban word for what in English is called a "river mouth", which is "river anus". Why? Because the anus is the exit point, and a river exits into the sea through its so-called "mouth". To a (hypothetical) native Lojbani, "river mouth" would probably be very confusing. -- John Cowan sharing account for now e'osai ko sarji la lojban. ______________________________________________________________________ >From ucleaar@ucl.ac.uk Fri Feb 18 19:21:13 1994 From: ucleaar Message-Id: <66062.9402181921@link-1.ts.bcc.ac.uk> To: conlang@diku.dk Subject: Re: in defense of natlang-style compounding Date: Fri, 18 Feb 94 19:21:13 +0000 X-Charset: LATIN1 X-Char-Esc: 29 Rick H: > 2. inefficient > "It" requires the insertion of extra phonemes into a compound in order > to indicate the relationship. Thus "it" lengthens words and decreases > efficiency, while providing little or no benefit in return; akin to > flexion of adjectives to indicate case, or flexion of verbs to indicate > person and number. Such flexion of adjectives & verbs has its uses. With adjectives, it allows you to identify the adjective's head. With verbs, it helps to identify the subject (and object if you also have object agreement). This takes some load off other areas, e.g. inflectional morphology of nouns, word order, etc. But I agree that the cost outweighs the gain usually. I can, however, imagine a language with, say, 30 genders, which inflects (or adds clitics to) the verb according to the genders of its complements. This wd make it easier to recognize which complement was which (& in fact complements cd often be omitted if, once you knew the gender, they were recoverable pragmatically). With 3 types of complement & 30 genders you'd need 90 clitics, but this wd buy you free word order among the elements of a clause with a 1 in 15 chance (I think) of ambiguity (assuming each gender has equal likelihood of occurring). ---- And ______________________________________________________________________ >From ucleaar@ucl.ac.uk Fri Feb 18 19:54:38 1994 From: ucleaar Message-Id: <48954.9402181954@link-1.ts.bcc.ac.uk> To: conlang@diku.dk Subject: Re: combining nouns via case; primitives Date: Fri, 18 Feb 94 19:54:38 +0000 X-Charset: LATIN1 X-Char-Esc: 29 Lojbab writes: > Amending that slightly. Our resident logician, John Parks-Clifford, specifi- > cally researched case theory as it stood in 1988 or so when we were doing > that par tof the redesign. His conclusion was that the leading theorists > were of the opinion that, while there was a possibility for a deep structure > theory of cases, that theory allowed for an effectively open set of cases. If predicates have a limited number of arguments, a limited number of cases will suffice to disambiguate these arguments. In fact the cases cd be semantically vacuous - just case-1, case-2, case-3. My conlang has 3 cases, but an indefinitely large and open class of predicate words. My cases are like 'nominative' and 'accusative', i.e. with some but not much semantic content. But the recent discussion on this list has also included cases like 'partitive', based on the PART-OF relation. This, I suggest, is a predicate. So for 'finger nail' instead of having 'finger' marked for partitive case, it shd be marked as 'accusative' (or whatever) argument of 'part-of'. fingernail: part-of(nail,finger) I feel that this answers the debate on whether there is (or should be) an open or closed number of 'cases' to link the components of compounds. The answer is that c.3 cases and an open class of predicates may be used to like the parts of compounds. A compound like 'foxhunting', where 'fox' is an argument of 'hunting' wd be permissible - as hunting(_,fox) - but not one like fingernail, where 'finger' and 'nail' aren't predicates. Fingernail would be something like 'nail-that-is-part-of-finger'. Just to make myself less unclear than I've probably been so far, I intend the above to be a resolution to the current debate. I hope that the suggestion would satisfy opponents of imprecise compounds, given the argument of the advocates of imprecision that the class of possible relations between components must be open. ----- And ______________________________________________________________________ >From robin@extro.ucc.su.OZ.AU Wed Feb 23 10:40:27 1994 From: Robin F Gaskell Received: from localhost (robin@localhost) by extro.ucc.su.OZ.AU (8.6.5/8.6.5) id XAA12858 for conlang@diku.dk; Tue, 22 Feb 1994 23:40:27 +1100 Date: Tue, 22 Feb 1994 23:40:27 +1100 Message-Id: <199402221240.XAA12858@extro.ucc.su.OZ.AU> To: conlang@diku.dk Subject: re: Syntax Analysis X-Charset: LATIN1 X-Char-Esc: 29 #From: robin@extro.ucc.su.oz.au (Robin Gaskell) 17 Feb 94 #To: conlang@diku.dk (Conlang Mail List) #Subject: re: Syntax Analysis Hi Conlangers, Well, on last count, the response to my piece on GAS - a means of clarifying the syntax of an uninflected language - was one. Perhaps the idea did not refer to recent work done by top researchers; or possibly, it duplicates work already done by AI workers, who are hot on the trail of the `meaning representation system.' As far as I know, it is an original ... and a Conlang first. >> Mark Shoulsen comments > Hmmm. Interesting method you have there, Robin. Is it really any better > than actually using the words instead of symbols? e.g. instead of > _. \! @._, maybe ? > OK, it certainly takes much more typing, but it's easier to read. It's > also language-dependent (whatever language you use for the words), and > could be less useful for marking sentences for analysis (as in _the .cat > \!sat @on the .mat_), though really not too badly: any computer worth its > salt could be perfectly capable of pulling out the information in the <>'s > and treating it separately. Though similarly, it could use the symbolic > method and translate to the format (in any of several > conventions/languages) for debugging output by the human operator. Thankyou, Mark for your feedback. Agreed, words are the readily readable medium. However, I wished to reach one level of abstraction beyond words, for at least two reasons: firstly, I had in mind the Chinese-speaker - learning Glosa - who wanted neither to learn English, nor to use a specially coined set of Chinese characters, to follow the mechanics of of a piece of Glosa prose; and, secondly, I preferred the elegance and economy of one symbol for one syntactic function. The reason, for wishing to keep the text separate from the syntactic coding, was more related to IT. While languages, which have inflectional grammar, do embody some of their grammatic functions in the body of their texts, their word orders still presents some problems to the programmer/parser-writer, and need their own analyses. Why not, I thought, make a clean break with this mixed system, and ... for the new `syntax-based' Glosa ... keep the text and gramatical analysis apart. I imagined that the text, unchanged, would hold the semantic content of a document, while the syntactic analysis, extracted from it, would act as a guide to the functional relationships between the elements of the text. As you point out, the information, of your preferred mode of syntactic analysis, can be extracted by computer. For the same reason, the _. \! @._ code can also be generated by computer. This GAS code would be much more economical of computer data storage memory; and, for purposes of visual monitoring, it would place less of a strain on the eyes of the person doing this monitoring - once they had learnt the forty odd codes. > You should realize, of course, that your method is quite limited to > languages that resemble English and Glosa. Not all languages have > participles, linking verbs, etc., and not all languages can have their > tense system so simply expressed as > past/present/continuous/conditional/now. Some require perfectives, etc. > Words like "gerund" and "passive" don't have any meaning in many languages, > but instead other constructs which you don't treat do. This may not > matter, but it will limit which languages can use this system. What you say is true. There may be some way of modifying this `syntax analysis' method to universalise it; however, that was not my point in intuiting GAS. The quite legitemate `syntax-based' system of grammar did not seem to have a suitable method for its analysis: I simply presented us with one. The historical grammars of the "unplanned languages" have been the subject of endless hours of enjoyable work by scholars, and the headache of parser-writers. And now, my philosophy is showing: in some ``Brave New World'' people will wish to communicate honestly; quite frankly, I see the Babel Syndrome as having reached the proportions of a social disease in need of treatment (or is it `intervention'); if this diagnosis is correct, then the IAL hypothesis ought to be the subject of serious research. At present, the best the world can offer is institutionalised linguists saying the study of "Interlanguage" reveals Esperanto to be the only legitemate object of their scholarship ... and we also have Conlang. Things are slow, on the Language Reform front. Despite this apparent inertia at the organisational level, I predict that there is a confrontation brewing. It will be between establishment figures, who, for reasons of national pride and personal position, support the present obfuscation, which gives them their status - and the under-current of humanity dissatisfied with an apparent lack of results in `the world language situation' area. A brief explanation of my philosophical outburst, and an analogy, might be in order. We Antipodeans still retain some of the intolerance with establishment that led our forefathers to seek a "fair go" in the new colony. The analogy: p 132, THE SUN-HERALD (Sydney), Feb. 20 1994. "In April, the World Congress on Cancer in Sydney will draw together nearly 40 scientists and clinicians from around the world to discuss their research into non- traditional cancer treatments. It is the first gathering of its kind, but the congress has had little support from the medical establishment, according to organiser Jennie Burke, director of the pathology centre Australian Biologics." When a friend of Jennie's went to Mexico for the successful cure of a cancer, that Australian doctors said was inoperable, Jennie decided it was time for a change. There is an air of dissatisfaction with establishment, about. Maybe it will gust through linguistics, soon, too. > Also, I note that you don't mark direct objects. This will limit your > method's usefulness to coding only those languages which mark case by > word-order, and that in the same order as English/Glosa. Coding "a fire I > saw" would not help us work out whether this was an OSV language (or more > likely, a SVO language with the O transposed to the front for emphasis), an > SOV language (very common) or a free-order language with case-markings. > This may or may not matter, depending on the purpose this code is intended > for (which I may not understand fully). Perhaps I have answered all this, above. Maybe I did not clarify the purpose of the GAS code in my earlier post. The code is designed to cover the area of languages with syntax-based grammar. It could be necessary to build-in the marking of direct objects; also, for different languages, it would probably prove helpful to declare - in a preface - the dominant order of the parts. Glosa being SVO (with optional Passive), and the one syntax- based language I am working on at present, the GAS system was fairly dedicated research, applying directly to it. English, with half of its grammar syntax-based, and the other half only minimally inflected, does submit fairly well to the GAS system. To apply syntax analysis to Esperanto, on the other hand, would certainly require some introductory declarative statements; and, owing to Esperanto's syntactic flexibility, would probably call for the addition of one or two further syntatctic categories, e.g. : might be used for the direct object. > It looks like a nice thing to run through your mind now and then to > see just what the language is doing, sort of like a computer-codable > version of sentence-diagramming (remember that?) Although I have not seen "sentence diagramming," I think you must have the idea. The GAS system is specifically designed to be computer-codable, with the twist that the code, being ASCII, is directly accessible via the computer keyboard, and is also readily networked. ---------------------------------------------------------------- As an appendix, I will repeat the first draft of the code:- The gASCII Analysed Syntax (GAS) system (Alpha Test ver.) ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Basic . ! > @ $ substantive action modifier space, time logical (noun) (verb) (adjective, preposition preposition adverb) Tense / \ ~ ^ | future past continuous conditional now Modifiers # % > v number quantity quality auxiliary (countable) (measurable) (property) verb Conjunctions + & joins words, phrases structural: joins clauses Functions x t = < <` location time equals, like verb is participle X proper noun as , similar passive 0 ? - , ; negative: un- general joining pronoun pronoun no, not, never question concepts personal impersonal nothing (compounds) People o s ' `. other self possessive gerund O proper noun S name of 1st person Specific ?o ?. ?! ?x ?t questions who what why where when Clauses ( ) { } [ ] " " adjectival noun adverbial parenthesis or quotation Non-literal : * * _ _ language metaphor or idiom start end other n-l term of sentence >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<< Examples of GAS in application 1. The cat sat on the mat. = U felis pa sed epi u tape. _. \! @._ 2. While three fat boys sat by the river bank, and ate jam sandwiches, their sisters stole their bicycles. _[t #>o \! @.-. & \! >.] ,'o \! ,'._ Tem tri paki ju-an pa sed proxi u ripa, e pa fago plu konfekti pani, mu plu fe-sibi pa klepto mu plu bi-rota. <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> Cheers, Robin ______________________________________________________________________ >From Edmund.Grimley-Evans@cl.cam.ac.uk Wed Feb 23 17:00:04 1994 Date: Wed, 23 Feb 1994 17:00:04 GMT From: Edmund.Grimley-Evans@cl.cam.ac.uk Message-Id: <9402231700.AA22665@nene.cl.cam.ac.uk> To: conlang@diku.dk Cc: "Edmund.Grimley-Evans" Subject: Glosa/Esperanto KOSMO GLOSA KOSMA GLOSO Panto-pe panto-lo ^Ciu homo ^cie ajn Pote uti Glosa; povas uzi Gloson, Plu panto-landa civi de ^ciu lando anoj, De panto nati-lingua. de ^ciu naski^glingvo. In panto domi, panto-lo En ^ciu domo ie ajn, Ofici e fabrika. en oficej', fabriko, Panto-pe dic; Glosa es ^ciu diras: Estas Glos' pro Gaia komunika. por Tera komuniko. Id dona interese ^Gi donas intereson a panto bon konversa al konversacio. Poli civi dice id ^Ciu ^statano parolas ^gin In Afika in Asia. en Afriko kaj Azio. In trena, navi, aeroplan En trajno, ^sipo, aviadil', Panto-speci vaga. dum ^cia voja^gado, Pe audi Glosa panto-lo Gloson oni a^udas ^cie: Id eko peri Gaia. ^Gi hejmas ^cirka^u Tero. Gru ad ali puta-me Ta^uga por iu komputil', poesi, e musika. poemoj, kaj muziko. Panto tema, panto-lo ^Ciun temon, ^cie ajn, Glosa don service. Gloso povas helpi. Dice, lekto, audi id Parolu, legu, a^udu ^gin tem sporta e relaxa, dum sporto kaj distri^go. Glosa es u maxi bon Gloso estas pleja bon' A fluvi, bun e saxa. ^ce rok', river' kaj monto. Glosa es u nece-ra, Gloso estas necesa^j', Un universa lingua. la lingv' universala. Fu doci sani panto-lo Instru' estonta pri sanec' E paci e eduka. kaj paco kaj eduko. Glosa pote proba Gloso povas provi ultra pan limita preter ^ciu limo, stop u lingua frustra la lingvan frustron ^san^gi Sti kosmo komunika. al Kosma Komuniko. Wendy Ashby [translation by Edmund GRIMLEY EVANS] Would any proponant of Glosa like to reciprocate with a decent translation into Glosa of "La Espero"? I hope no one objects to this message. I joined the list only a few days ago, so I don't yet know precisely what is acceptable here. I joined after seeing conlang in the list of publicly available distribution lists in the news group news.lists; conlang was a new addition to that list, though I realise that the list itself is not new. It was a good idea to get it listed, in my opinion. Edmund ______________________________________________________________________ >From lojbab@access.digex.net Wed Feb 23 12:10:49 1994 From: Logical Language Group Message-Id: <199402232210.AA03039@access2.digex.net> Subject: Re: combining nouns via case; primitives To: conlang@diku.dk Date: Wed, 23 Feb 1994 17:10:49 -0500 (EST) Cc: lojbab@access.digex.net (Logical Language Group) And Rosta writes: > But the recent discussion on this list > has also included cases like 'partitive', based on the PART-OF > relation. This, I suggest, is a predicate. So for 'finger nail' > instead of having 'finger' marked for partitive case, it shd be > marked as 'accusative' (or whatever) argument of 'part-of'. > fingernail: part-of(nail,finger) This is fine for simple compounds, where the relationship "between" is clearly a two-place predicate. But consider compounds like "tooth root" (or Hungarian "coat finger" (meaning "coat sleeve")). Here the relationship is something like: analogy(part-of(root,TREE), part-of(tooth_root, tooth). This is not a simple predicate, and involves the implicit thing-which-really-has a-root which I have marked "TREE" here. Given that there is a predicate "A is a part of B which is analogous to the C'th part of D", how do we mark which of "tooth" and "root" is being assigned to A, B, C, or D? This is what I meant by the notion that using the general predicate vocabulary to mark compounding relationships was recursive. We have two predicates to relate, pred1 and pred2. We mark the relationship between them as pred3. We then must mark the relationships between pred1-pred3 and pred3-pred2 as pred4 and pred5, respectively..... -- John Cowan sharing account for now e'osai ko sarji la lojban. ______________________________________________________________________ >From hrick@world.std.com Wed Feb 23 18:36:38 1994 Date: Wed, 23 Feb 1994 23:36:38 -0500 From: hrick@world.std.com (Rick Harrison) Message-Id: <199402240436.AA03499@world.std.com> To: conlang@diku.dk Subject: Vorlin update X-Charset: LATIN1 X-Char-Esc: 29 A couple of weeks ago, And Rosta said it's a shame we don't hear much about Vorlin anymore. Well, that was a very nice thing to say, and it inspired me to dig up the old computer files (untouched for nearly a year) and ponder the possibilities. Vorlin has some serious handicaps. Some compound words are hard to pronounce (/s/ followed by /sh/, /p/ followed by /b/, /sh/ followed by /zh/ etc). The initial premise, that a sufficient vocabulary can be built from CVC roots, might be erroneous; in the latter days I was experimenting with adding some CVCVC roots. (An a priori battery of CVC roots might be sufficient, but coming up with sufficiently distinct and memorable a posteriori CVC radicals is difficult.) And I no longer believe in mandatory tense indication for verbs or mandatory flexion of nouns to indicate number. Regrettably, I should have learned more about linguistics and conlangery before I published Vorlin. On the other hand, it was publishing Vorlin that got me connected to the quasi-community of people interested in a variety of constructed languages. The good news is that no-one has memorized the last version of Vorlin (as far as I know), so I can make changes without inconveniencing anyone. And indeed I have begun making changes. Expect something to be published around March 21st (Vorlin's 3rd birthday). ______________________________________________________________________ >From lojbab@access.digex.net Fri Feb 25 06:25:42 1994 From: Logical Language Group Message-Id: <199402251625.AA12718@access2.digex.net> Subject: Re: Vorlin update To: conlang@diku.dk Date: Fri, 25 Feb 1994 11:25:42 -0500 (EST) Cc: lojbab@access.digex.net (Logical Language Group) X-Charset: LATIN1 X-Char-Esc: 29 Rick Harrison writes: > Vorlin has some serious handicaps. Some compound words are hard > to pronounce (/s/ followed by /sh/, /p/ followed by /b/, /sh/ > followed by /zh/ etc). You could adopt the Lojban mechanism for dealing with difficult clusters: introduce the schwa, written "y", for separating them. This requires a definition of "difficult", so that words can have a single canonical form. The Lojban definition of a forbidden cluster is fourfold: 1) duplicates are forbidden: *kk, *rr, *ss. 2) voiced-unvoiced pairs are forbidden unless one consonant is l, m, n, r: *bf, br, *zs, zn. 3) clusters where both members belong to {/sh/, /zh/, /s/, /z/} are forbidden. 4) the special cases cx, kx, xc, xk (which are hard to articulate) and mz (which is easily confused with nz). Others might want to add restrictions based on homorganicity or the lack of it. > The initial premise, that a sufficient > vocabulary can be built from CVC roots, might be erroneous; in > the latter days I was experimenting with adding some CVCVC roots. This sounds useful. In particular, you will need to be able to borrow long forms for representing the multiplicity of living things, cultures, specialized foods, and other such referents. > (An a priori battery of CVC roots might be sufficient, but coming up > with sufficiently distinct and memorable a posteriori CVC radicals is > difficult.) An excellent point. The short radicals of -gua!spi (C^nV^n, where V includes [aeioumnlr]) are technically a posteriori, being derived from English, Latin, and Chinese, but they are not very mnemonic. > And I no longer believe in mandatory tense indication for > verbs or mandatory flexion of nouns to indicate number. Good! -- John Cowan sharing account for now e'osai ko sarji la lojban. ______________________________________________________________________ >From riddle@is.rice.edu Sat Feb 26 05:26:34 1994 From: riddle@is.rice.edu (Prentiss Riddle) Message-Id: <9402261726.AA14116@is.rice.edu> Subject: Volapuk Renaissance To: CONLANG@diku.dk (ConLang list) Date: Sat, 26 Feb 1994 11:26:34 -0600 (CST) X-Www-Page: http://is.rice.edu/~riddle/index.html X-Charset: LATIN1 X-Char-Esc: 29 Writing in Linguica #22, Michael Helsem said the following: > Books on, in, or about Volapuk are extremely hard to find. > Basically, nothing has been reprinted since de Jong's German-Volapuk > Volapuk-German dictionary (1931), & for English you have to go back > to Seret (1887). Even the Volapuk Center in England (Brian Bishop) > can only xox the copies they have (@ 10p a sheet). I only just > received, from a college in Virginia via Interlibrary loan, the > 106-year-old book, which i plan to copy (very gingerly) in the days > to come... It's been checked out 5 times: in 1932, 1959, 1968, 1986, > 1987 & 1989 (is that a renaissance or what?). > > Ralph Midgley, who does the newsletter (24 Staniwell Rise, > Scunthorpe, DN 17 1TF United Kingdom) has put together a condensed > grammar & English-Volapuk only wordlist (but thoughtfully includes > the Volapuk hymn...), & Joseph Biddulph sells the grammar portion of > Seret (the first 59 pages) for 6.5 pounds: 32 Stryd Ebeneser, > Pontypridd, CF37 5PB, Wales. Biddulph's is the kind of small press i > like: not many books, & they're not very big, but where else can you > find out about Frisian, Visigothic, Mandingo, Bushman or Twi? (Also > Basque & Gujarati.) Midgley sent me the wordlist for free (since i > am what is probably the first original poetry in Vp. since before > WWI) but a donation is suggested & it has a list of the affixes, > something oddly missing from the other books i have. In answer to my > inquiry, Joseph Biddulph says he might consider a reprint of the rest > if there were a definite demand for it -- i don't know if that means > one, ten, or a hundred or more letters -- if i had access to the > Conlang BBS [sic] on Internet, i would do some agitating, but... There you have it. If you would like to see a reprinting of a primary source on Volapuk, drop a line to Joseph Biddulph. If you want to reach Michael Helsem, drop him a line at 1031 Dewitt, Dallas, TX 75224-2651. -- Prentiss Riddle ("aprendiz de todo, maestro de nada") riddle@rice.edu -- Opinions expressed are not necessarily those of my employer. ______________________________________________________________________ >From ucleaar@ucl.ac.uk Sun Feb 27 12:55:48 1994 From: ucleaar Message-Id: <24097.9402271255@link-1.ts.bcc.ac.uk> To: conlang@diku.dk Subject: Re: Volapuk Renaissance Date: Sun, 27 Feb 94 12:55:48 +0000 X-Charset: LATIN1 X-Char-Esc: 29 Prentiss Riddle quoting Michael Helsem about two British Volapuekists reminds me of a feature on British TV a few months ago about Volapuek. The programme was called Eurotrash, celebrating features of European culture that are kitsch, lewd or risible, all introduced by Jean Paul Gautier & Antoine de Caunes bickering in franglais. One item presented two English men who, it was claimed, were the only Volapuekists in Britain but had never met before being introduced by the programme. They were filmed conversing in V. & singing the V. hymn. I found them rather charming & dignified, and a rather peculiar inclusion in a magazine programme most of whose features were about avantgarde pomo Luxembourgeois pornographers. ----- And ______________________________________________________________________ >From dasher@netcom.com Sun Feb 27 15:17:33 1994 Date: Sun, 27 Feb 1994 23:17:33 -0800 From: dasher@netcom.com (Anton Sherwood) Message-Id: <199402280717.XAA15865@mail.netcom.com> To: conlang@diku.dk Subject: Biddulph's publications X-Charset: LATIN1 X-Char-Esc: 29 Announcement found in the back of a booklet on Visigothic: Two Serial Publications-- Edited by Joseph Biddulph: _Xenododo_: _A Miscellany of Ancient and Exotic Tongues_. Irregular. 24-28 concise pages. ISSN 0957-5375. UKL2.00 per part. There are absolutely no boundaries in the studies proposed by _Xenododo_ - living and extinct languages and dialects,European, Asiatic,Oceanic,American,African linguistics, Scots,Creole,English and Celtic dialect,a little anthropology - with useful addresses, and an invitation to readers to participate - popular and academic items from a very wide field.Published in basic format in English and any other European language, and obtainable as a series or as separate parts. Write for details today! _Hrafnhoh_ ISSN 0952-3294 Irregular. 24-28 concise pages. A stimulating mixture of heraldry and philo- sophy,genealogy and poetry,literature and antiquity,placenames and reviews,etc.,etc. Write for details to: Languages Information Centre, 32 Stry^d Ebeneser, Pontypridd CF37 5PB, Wales/Cymru. *\\* Anton Ubi scriptum?