archives of the CONLANG mailing list ------------------------------------ From: hrick%genesis.uucp@gte.com (Rick Harrison) Subject: ostensible opposites Date: Mon, 6 Jul 92 22:40:27 Being a compulsive list-maker, I was recently toying with the idea of trying to list all the possibilities which a language designer must (consciously or unconsciously) consider when inventing a language: which phonemes to include, whether to use adpositions or noun declensions or a combination, etc. (And I've added this project to my list of lists to make.) One thing that crossed my mind was the various ways in which languages express qualities that are presumably opposites. (This topic was briefly discussed on this list almost a year ago.) There are 3 systems that I can think of: 1) The words representing qualities that are ostensibly opposites show no apparent relationship, e.g.: Middle Vorlin: bona = good, mala = bad English: broad = broad, narrow = narrow Poetic Esperanto: varma = toasty, frida = chilly 2) Use of an affix which indicates "direct opposite of...", e.g.: Esperanto: nova = new, malnova = old Old Vorlin: bona = good, kanbona = bad 3) Use of affixes which I'll call intensive, normative, and attenuative (I don't know if there is existing linguistic jargon to describe these, so I invented my own terms). I'm not real familiar with any lang's that use this system, so we'll make up a hypothetical conlang in which "makro-" is intensive, "meso-" is normative, "mini-" is attentuative, "pesad" means "weight (relative measure of heaviness or lightness)," and the suffix "-ik" indicates an adjective. This produces... makro-pesadik = heavy meso-pesadik = medium-weight mini-pesadik = light-weight My questions, to anyone who cares to comment, are: Do any natlangs use method #3? How about conlangs? Which system do you favor, and why? Might it be desirable for a conlang to use a _mixture_ of these systems, for example having separate words for high-frequency opposites like "good" and "bad" and using method 2 or 3 for less-frequently-discussed qualities? ===================================================================== Rick Harrison hrick@genesis.nred.ma.us = gte.com!hrick%genesis.uucp Journal of Planned Languages, Box 54-7014, Orlando FL 32854-7014, USA Date: Thu, 9 Jul 92 12:34:08 MDT From: mnu@inel.gov (Rick Morneau) Subject: Re: ostensible opposites Howdy conlangers! Rick Harrison asks some interesting questions about opposites, which are really questions about the interaction between morphology and lexical semantics - two of my favorite subjects. Linguists have written quite a lot on the subject. For a highly readable reference, I recommend "Lexical Semantics" by D.A. Cruse, Cambridge University Press, 1986. The book contains three chapters just on opposites. Here's a partial breakdown of the kinds of opposites you can run into: Complementaries: mutually exclusive - no gray area, such as true/false, dead/alive, hit/miss Antonyms: gradable, such as long/short, good/bad, hot/cold. This group has several sub-groups. Directional opposites: north/south, up/down, forwards/backwards Antipodals: cellar/attic, head/toe, full/empty Counterparts: male/female, ridge/groove, heaven/hell Reversives: rise/fall, enter/leave, tie/untie Converses: before/after, above/below, in front of/behind Relational: doctor/patient, predator/prey, parent/child and much more. Fortunately or unfortunately, I've never heard of a natural language that marks its words consistently and equally for polarity. When it occurs at all, there is always an imbalance, as with the English "un-" (For example, "marked" is unmarked, while "unmarked" is marked. No smiley intended or required.) However, both natural and artificial languages mark their words for many other things, and I see no reason why we can't extend this idea to cover polarity. For example, in many Malayo-Polynesian languages (Indonesian, Tagalog, et al.) verbs are derived from primitives to mark their transitivity and whether they are active or passive. Or, consider Iroquoian langauges, such as Mohawk and Cherokee, which mark their verbs for the semantic roles agent and patient played by the arguments. As for nouns, consider the Bantu langauges which mark for class (which is often semantically relevent). What if we extended these morphological derivations to indicate their polarity by doubling the number of marker morphemes? In other words, where you may now have one morpheme to mark a specific feature, instead have two: one for positive and one for negative. For example, let's say your conlang marks its verbs for transitivity, like this: ke- = positive, transitive (X does something to Y) bu- = negative, transitive " di- = positive, intransitive (something happens to X) sa- = negative, intransitive " If the root for "free, non-captive" is "booga", then we could create the following words: kebooga = to free, to release bubooga = to capture dibooga = to escape sabooga = to surrender This approach is okay for a crude first pass, but it is unsatisfying as a real solution because transitivity is a *syntactic* phenomenon. I feel that a *semantic* solution based on thematic relations would be much more flexible, and I'll have LOTS to say on this in the not-too-distant future. Also, this solution is too simplistic, because it fails to deal adequately with gradeable concepts, like "hot", "warm", "lukewarm", "ambient", "cool", "cold", "frigid". For these, you could start with derived primitives to handle basic (+), (0) and (-) concepts, and then add fine-tuning morphemes to cover the cases in between. For example: positive (+) primitives: CVNC, where N is a nasal, homorganic with the C that follows it. neutral (0) primitives: CVLC, where L is a liquid, such as /l/ or /r/ negative (-) primitives: CVSC, where S is either /s/ or /z/, agreeing in voicing with the C that follows it. Next, add the fine-tuning morphemes: ki- = to move in a positive direction no- = to move in a negative direction So, to put it all together, let "pend" = "hot", "peld" = "ambient", and "pezd" = "cold". With the fine-tuning morphemes, you can now get: kipend = scalding, red-hot pend = hot nopend = warm, not too hot kipeld = lukewarm, tepid peld = ambient, room-temperature nopeld = slightly cool kipezd = cool pezd = cold nopezd = frigid You can, of course, add more fine-tuning morphemes to handle finer distinctions or more extreme ones, such as "super hot" and "super cold". The only problem I have with all of this is that it is definitely unnatural, and I would normally be reluctant to include a feature in a conlang that could potentially go against some linguistic universal. However, I doubt if this is a serious problem, since linguistic purists could always regard the combination of a "primitive + polarity marker" as a single morpheme. I will say no more about this because I am currently working on an essay that goes into considerably more detail on the whole topic of lexical semantics for conlangs, and I prefer to say it all there rather than here. I will make it available to members of this list when it is ready. Regards, Rick -- *=*= Disclaimer: The INEL does not speak for me and vice versa =*=* = Rick Morneau Idaho National Engineering Laboratory = * mnu@inel.gov Idaho Falls, Idaho 83415, USA * =*=*=*=*=*=*=*=*=*=*= NeXT Mail accepted here! *=*=*=*=*=*=*=*=*=*= >From maxwell Fri Jul 10 16:01:32 1992 Subject: Dulic^enko's book Date: Fri, 10 Jul 92 16:01:32 MET DST About two months ago, there were several postings about a book in Russian about CLs by Dulic^enko of Estonia. A week or two ago, I received the April(!) issue of *Esperanto* with a review of this book by Detlev Blanke (Germany). I haven't bothered translating the whole review, but I thought a few excerpts (my translations) might be worth the trouble. Quite a few familiar names come up. My comments are enclosed in [], comments by Blanke, years, etc. in (). .... Dulic^enko is known among other things as the editor of the series *Tartuensis*, of which the 7th volume has already appeared (1990). ...... Who before D's work could realistically answer the question about how many planned language projects there are? Couturat/Leau (1903/1907) estimated 75, Gue'rard (1922) mentions 106, Stojan (1929) lists 350, Drezen (1931) 370. Monnerot-Dumaine (1962) with 382 didn't progress much. And also the text of Janton (1973) with 500 doesn't have a very firm basis. The most realistic guess(?) was by Mario Pei en 1969, who indicated 1000! All those mentioned were or are informed specialists on the subject. Let us pardon the linguists who in this regard are not well-informed: Akulenko (1971), who assumed 200, and Haarmann (1973), who in a very superficial study suggests only 20 projects in the 19th and 20th century. ..... The manuscript was ready in 1973 and had to wait a long time for publication. Because of this, the material is not at all up-to-date. According to the author's records there had appeared 912 systems up to that time. So maybe there is more motivation now (than Pei had) for estimating the total number as being between 960 and 980. Unfortunately, there is a lack of journals (like *International Language Review* or *Union*, later *Planlingvistiko*), which not only provide a record or new projects but also stimulate debutants in project construction. Maybe *Vidpuni* (Rick Harrison [snail-mail address added]), which has been appearing since 1991, can become such a review. ..... In 1887-1973, that is, after Esperanto until the editorial deadline, there appeared more than 600 projects, almost 70% of all recorded systems. There follows a table of the project-names .... The author names reveals to us a few particularly productive creators: Rene' de Saussure with 10, Molee with 8, Bond, Lott, Weferling and Weisbart with 6 each, Hendersen and Lallemand with 4 each. Also the list of countries presents surprises. One plan-language author came from each of the following countries: Guatemala, Panama, Philippines, and south Vietnam. The leading country is France (151 projects), West Germany (including West Berlin, no projects appeared in East Germany?) follows with 136 projects. Third place is occupied by USA (98) than Great Britain (84) and Soviet Union (82). .... Resume's given in Estonian, French, German, and Esperanto (but not in English, why not?) helps those who don't know Russian to orient themselves in this valuable treasury. ...... The number of volumes published is only 1000. It would be very desirable to translate the entire work into Esperanto. At the same time, it needs to be updated, since a whole bunch of new plants have recently appeared in the planned language garden, among them very interesting ones, like *Glosa*(R.Ashby/W.Clark)1978, [for most of the rest of the list I give only the name of the language] Sprache 2000, ... Vikto...Adli...Uropi...Unitario...Vorlin, Rick Harrison, 1990(?). Will Dulic^enko add an appendix? Let's hope so. Dan Date: 12 Jul 92 01:57:10 EDT From: Don HARLOW <72627.2647@CompuServe.COM> Subject: JPL15 To: Conlang >INTERNET:conlang@buphy.bu.edu Dato: 920711 Kudos to Rick Harrison for Journal of Planned Languages 15, two copies of which arrived here yesterday. I found the article by X. Rayburn on Elisabeth Wainscott's UNI very interesting. Several years ago, the ubiquitous Bernard Golden published an article in "Latinamerika Memuaro" from Colombia on four woman conlang designers; Wainscott was one of them, if I remember correctly (a second was Wendy Ashby of Glosa, and I don't remember who the other two were; if I find the article, I'll send it along). Golden's article, however, concentrated more on the personalities and activities of the individuals involved than on the languages themselves. News and notes were also interesting, though nothing (so far as I could tell) that someone in this forum would not already know about. Few of the "publications received" have been mentioned here, though. The only sour note, to me, was Mark Tierisch's review of Yaguello's "Lunatic Lovers of Language". Not having a copy of the book, I can't either agree or disagree with his conclusions (seems to me that I would probably agree with most of them); but his own political opinions seem to play far too large a role in the review, and he himself makes a few unprovable comments. "The belief that all human languages have a single, common origin was eventually discredited by scientific investigation" -- with a time machine, perhaps? "... the myth of the lingua adamica resurfaced in a pseudo-scientific 'reconstruction' of a world root-language which was invented in the USSR before communism was squashed there and which was immediately, unquestioningly embraced by socialists in academia and elsewhere"; "Her guesswork on this matter is marred by leftist 'political correctness;' the possibility that there might be genuine, innate differences between the minds of men and women never occurs to her, because this possibility does not mesh with her politics." One paragraph particularly bothered me ... probably because I've written much the same thing, in almost the same words, and yet I find it very troublesome. "Part 4 of Yaguello's opus is entitled 'In defence of natural languages.' I find it amazing that anyone would take such a position." Tierisch then proceeds to show (at least in certain fields) the superiority of "artificial" solutions over "natural" non-solutions. On this I agree with him. But I can no longer find it amazing that anyone would support a natural language as opposed to an artificial language. The vast majority of humans, had they grown up in caves, would find it very difficult to move into houses; and language has a much tighter hold on the human spirit than dwelling-styles do. Anybody out there who can read Esperanto should get a copy of Piron's paper on "Psychological Reactions to Esperanto". The conclusions of Piron, who is a practicing psychiatrist-counsellor in Switzerland, can, I think be generalized to any conlang which might find itself in the same (even remotely) "threatening" position as Esperanto. The reader will, I think, understand Yaguello's opinion (which is shared by many, many other people). Date: Tue, 14 Jul 92 07:22:12 -0700 From: chalmers@violet.berkeley.edu (John H. Chalmers Jr.) To: conlang@buphy.bu.edu Subject: ostensible opposites, gradable series I was reminded by Rick Morneau's examples of a gradable polarity (pend, peld, pezd with the meanings hot, room- temperature, and cold) of the "apophonic" sequences postulated to be a feature of the original pan-human ("proto-world") language. This theory was developed in some detail by Roger Wescott in a report published in the late 70's of a conference on language origins, though as I recall, he credited the idea to Morris Swadesh and Mary LeCron Foster (?). Unfortunately, I am in San Diego and do not have my references handy. If anyone is still interested, I will look up the refs after I return to Houston later this month. Wescott suggested that the terminal phonemes l, n, and r formed a series with attenuative, normative, and augmentative values. He gave alleged examples from English, but my recollection is that they could be explained by borrowings from various dialects and languages. In any case, the presentation was rather unconvincing. -- John. From: cbmvax!snark.thyrsus.com!cowan@uunet.UU.NET (John Cowan) Subject: Re: Drumtalk (was: Languages of colour and sound?) Date: Thu, 16 Jul 92 11:28:55 EDT Ken Beesley writes: > Some tonal languages of Africa can similarly be > transmitted by beating drums tuned to different notes (the "Bush Telegraph"). > Some have suggested that these transmission systems, being limited to tones, > require that you transmit fairly conventional messages to insure understanding. > Others claim that arbitrary messages can be transmitted. African drumming is > probably the most sophisticated in the world, often involving long training > from childhood within traditional religious societies, and I suspect that these > drummers can express a lot more than just tonal distinctions. My understanding (but I am no expert) is that >all< these claims are true, representing simplifications from different angles of the whole truth. The key is (as they say on talk.bizarre) "Volume volume volume!" Since the drumtalk indeed carries only tonal information, it is necessary to expand each sememe of the spoken message into a longer, fancier spoken form whose tonal pattern is less ambiguous. For many sememes, standard expansions exist and are systematically taught; for others, an expansion must be created on the fly by the drummer. An English analogue is the rhythmic (non-tonal) pattern "daaah dah-dah daah daah (pause) dah dah" which is heard as "Shave and a haircut, two bits". Obviously there is not enough information in the drumbeats for this message to be reconstructed, but it is obvious to those with appropriate knowledge, and could be used in context to represent some sememe or other, possibly "shave" or "barber". Note that the tonal phonemes of most African languages are based on tone levels rather than the tone contours of most Southeast Asian languages, so pitched drums are sufficient to represent them. -- John Cowan cowan@snark.thyrsus.com ...!uunet!cbmvax!snark!cowan e'osai ko sarji la lojban. From: And Rosta Subject: Damin Date: Tue, 21 Jul 92 23:16:16 +0100 R.M.W.Dixon's (1980?) _The languages of Australia_ (Cambridge UP) &, more recently, & very eloquently, Kenneth Hale's "Language endangerment and the human value of linguistic diversity" in _Language_ 68.1 (March 1992), pp35-41, discuss the language Damin, spoken by initiated men of the Lardil linguistic community. Because Damin differs phonologically from Lardil, all other Australian lgs, and, as far as is known (of countless languages we of course know nothing), from all other human languages (Damin has an ingressive lateral fricative and also I sound that I guess, from the phonetic descriptions, may be what is called in England, at least, "a raspberry"), it is conjectured that Damin was invented. Lardil folklore does indeed hold that one of their ancestors invented it. But, as Hale points out, invented languages are nearly always phonologically close to the language of their inventors. [The unpublished conlang Sta, whose phonology was presented on Conlang some months back, is another exception. Interestingly, Sta and Damin both have typologically rare nasalized (and only nasalized) clicks.] Damin uses the grammar of Lardil, but has a lexicon all of its own. This lexicon is very small, but Hale shows how it was nevertheless feasible. Hale's regrettably brief discussion is likely to be of interest to 'conlinguists'/glossopoeists. Damin was once a living language spoken in daily life, although not learnt natively. This suggests that the use of artificial (and perhaps straightforwardly *invented*) languages is not a twentieth century hobby for _foux de langages_ but is rather much more pervasive in human cultures. --- And. Date: 23 Jul 92 09:09:00 EDT From: "61510::GILSON" Subject: Final (I think) status report to Conlang group on the Voksigid project This functions both as the apparent final progress report on the Voksigid project to the conlang group and the apparent final communication to the Voksigid development group, a. k. a. the newlang group. If anyone, either by sending to the Voksigid group as a whole at newlang@buphy.bu.edu, or by emailing to me at brgilson@highlite.gotham.com, evinces an interest in reviving the Voksigid project, the possibility of continuing the effort still exists, but I am pessimistic about the chances. My observations are that the demise of the effort was not due to any impracti- cality of implementing the concept that Voksigid represented. I still feel that we might have, if a few non-linguistic happenings had been different, come out with something better than Lojban at the things Lojban is best at, while being easier to learn. Voksigid went through three phases: 1. In a preliminary organization phase, we eliminated two people whose ideas were so far removed from what the rest oif us had in mind that compromise was clearly impossible. One of them, I feel, could well have contributed a lot of useful ideas, but was so firmly wedded to an a priori vocabulary that we could not hold him. His loss was unfortunate because I think he knew a lot about some aspects of grammar that most of us did not know. The second of those would at least have given us some input from a person whose native language was not English, but it was very clear from early on that there was no compromise that could embrace both him and myself; on one occasion I went so far toward his proposal that it was making me ill to conceive of what was being done, and yet he was accusing me of being unwilling to compromise. With the departure of those two from the group, we were able to come up with some documents to define the language, and at that point it looked as if we were making progress. 2. At that point, we began filling in the details. We developed a vocabulary, fleshed out a few grammatical details, and I thought we'd soon have a language created. Then disaster struck. One of us, who had been the most prolific source of ideas in the first phase (and who, more than I, was the person whose structure as first proposed turned out to be the one that the final Voksigid resembled most closely) had to leave to devote full time to his dissertation. Another changed schools. I myself, for a while, was incommunicado because I lost the ability to receive e-mail at my work computer and needed to establish a new location. That led to phase 3: 3. We got to a point where a proposal would be made, and nobody would respond. It was clear that nothing more was going to happen. This is where we are now. I think we have to give Voksigid a decent burial. The defining documents are still, I assume, to remain on the PLS, and if someone sends a copy of Dave West's final version of the lexicon to the PLS archives, they will have a basis on which to proceed if anyone wants to take the role that Ashby and Clark took to Hogben's Interglossa. Observations: Apparently, for an experimental language, the only organization scheme that works is something like what happened in Lojban. One person (in that case, JCB) developed a language, and a group was set up only much later, but the group was much larger than we had, so that dropping out by 1 or 2 or 3 did not leave them so shorthanded that paralysis started whenever two people disagreed. The voting mechanisms I devised worked when we had 5 or 6 people; when we got down to 3 we were lost. They would have worked better if we had the 9 or 10 I'd originally envisioned. Lojban had one unfortunate experience, the split between JCB and lojbab that required them to construct a totally new vocabulary from scratch -- they'd be a year or two further along, I think, if that hadn't hap- pened. But (even though I can't read the language well enough to follow every- thing that goes on) I think the group works well. I'd hoped that we could do that kind of thing eventually. The only problem is getting to that point. A committee such as we had doesn't seem to work. I'm sorry it doesn't. John Ross and Jim Carter, at least, had useful ideas without which the language, if it had depended on me alone, would never have been as good as the one we were on the threshold of developing. Jim, in particular, even though he has his own creation (guaspi) and is also an active participant in the Lojban group as well, was able to grasp the spirit of Voksigid well enough that his sug- gestions were frequently right to the point, even though they had to be different from the way guaspi or Lojban would handle the same problem. I wish we had been able to get the same from the one person I mentioned earlier who was the first to get off the boat. Bruce Date: Mon, 3 Aug 1992 10:37:43 PDT From: Ken_Beesley.PARC@xerox.com Subject: [mnu@inel:gov:Xerox: Infixes] Bob LeChevalier writes: > > Lojban of course uses infixes, since its affix system used in word > making allows the affixes to be used in all positions in a word - > prefix, suffix, and infix. > Rick responds: >Although I'm too lazy to look it up, I believe your definition of >"infix" is too broad. An infix is a morpheme that appears within > another morpheme or which modifies the morpheme internally. Affixes >which are simply attached to a stem are either prefixes or suffixes, > even when preceded or followed by additional affixes. Rick is quite correct here. Rick's Arabic examples show a kind of infixing (semitic root and pattern interdigitation), but something like Tagalog is probably more typical. One form of the Tagalog verb "bili" is "bumili," involving infixation of 'um' after the first consonant (b-um-ili). Similarly "binili," with an infixed -in- after the first consonant. Another interesting productive process in this language is reduplication, "bibili," involving the duplication of the first syllable, whatever it might be. Even reduplication plus infixing is possible: bumibili. These processes are productive and convey predictable meanings. Infixation may appear odd to most speakers of Indo-European languages, but it appears to offer no identifiable difficulties in human processing. Various English "play languages" (the term often used in linguistics for Pig Latin, Double Dutch, Turkey Talk, Carney Talk, etc.) have similar infixing constructions, though obviously without much semantic content. In computer processing of morphology, Arabic interdigitation and Tagalog infixing have also been handled quite successfully in a couple of projects, but I wouldn't be tempted to use either process in a constructed language. Such languages are definitely harder to deal with, creating unusual challenges in lexicography and automatic morphological analysis. I would be quite interested in a copy of Rick's essay on applying Arabic morphology to the design of constructed languages as well as any other related messages. Ken Beesley beesley.parc@xerox.com Date: Tue, 4 Aug 1992 11:10:00 PDT From: Ken_Beesley.PARC@xerox.com Subject: Re: Do I always seem to disagree with Rick M? %^) lojbab writes: >The rafsi -jav- for the tanru-logical-connective-OR makes no sense except b >between two other components; hence I have trouble seeing it as other than a >an infix. No, Rick is spot on here. I'm afraid that your usage of the term "infix" is contrary to all standard linguistic terminology and is not a useful usage. The basic question in morphotactics is whether the language builds words by simply concatenating morphemes together, or by inserting one morpheme inside another single morpheme, or by interdigitating morphemes together. Your example with -jav- appearing "between two other components" is a simple case of concatenation. One should not confuse infixing with compounding, prefixing, suffixing, and the issue of whether a particular morpheme can be a stem. We should not confuse it with the idea that a particular morpheme can be or must be surrounded by other morphemes. The key point is that when the morphemes in a word appear one after another, like ducks in a row, you don't have infixing, just concatenation. A true infix, like the Tagalog -um-, appears not "between two other components" but actually tucked inside another single morpheme. Thus "bili" is a single morpheme, and "bumili" is a form of "bili" that has -um- infixed after the first consonant. The result in the final word is that the morpheme "bili" is phonetically and orthographically discontiguous. The Arabic 'alkaatibu ("the book (nominative)") exhibits both concatenation and a special kind of infixing often called interdigitation. The morphemes are 'al- the (prefix) ktb the triliteral root CaaCiC the nominal pattern, with 'C' indicating the slots where root radicals go -u definite, nominal (suffix) Like interdigitating fingers, the root and pattern lace together to form the stem kaatib. The result is that both the root and pattern morphemes are discontiguous in the final word. Then the prefixes and suffixes are added in the usual concatenative way. Infixing (and interdigitation) result in single morphemes being realized discontiguously. In practice, infixing is relatively hard for learners and computer programs to deal with, mostly because it requires them somehow to reassemble discontiguous morphemes for dictionary lookup. In comparison to infixing, the morphotactics of even the most fearsome concatenating systems are usually a piece of cake. Ken Beesley From shoulson Tue Aug 4 14:24:56 1992 EDT Subject: [mnu@inel:gov:Xerox: Infixes] Yes, infixing (of the Semitic style) does tend to break IE concepts. Yesterday, when I noticed that Rick H. was looking for natural languages as well as conlangs for his dictionary, I thought I might try my hand at doing Hebrew (probably not the best plan; for all that I can speak it, I'm not a native speaker). I decided against it in the second group, when I realized that there's no reasonable suffix/prefix/anything that works for "-able", apart from complex infixation/interdigitation of vowels, and even that has its irregularities. The pedantic could argue that Esperanto uses infixes, since what are called its "suffixes" really go before the part-of-speech marker (some pedant once took me to task for referring to the dreaded feminine suffix of E-o as "-in"; he maintained that I should write "-in-", since stuff follows it.) But you could (and probably should) also view it simply as successive suffixes, like Okrand's Klingon with its 5 classes of noun suffixes and its 9+1 classes of verb suffixes. Lojban's rafsi don't seem to have the infix nature to me. When I hear "infixes" regarding a language, I think of something more along the Tagalog lines, or the children's language that infixes some syllable (I heard it with "ab") before all vowel sounds. Lojban's affixes are more just short, combining forms of words, not affixes in the sense of little bags-on-the-side of words that happen to hang in the middle. ~mark Date: Tue, 4 Aug 1992 12:03:00 PDT From: Ken_Beesley.PARC@xerox.com Subject: Re: [mnu@inel:gov:Xerox: Infixes] Mark Shoulson writes: >The pedantic could argue that Esperanto uses infixes, since what are called >its "suffixes" really go before the part-of-speech marker (some pedant once >took me to task for referring to the dreaded feminine suffix of E-o as >"-in"; he maintained that I should write "-in-", since stuff follows it.) >But you could (and probably should) also view it simply as successive >suffixes, like Okrand's Klingon with its 5 classes of noun suffixes and its >9+1 classes of verb suffixes. >Lojban's rafsi don't seem to have the infix nature to me. Mark, You're quite right. The -in- or -in (ignore pedants) of Esperanto is a garden variety agglutinating suffix that just happens to require another morpheme to follow. Same for the -ad- in verbs like "diradis". Still just suffixes. Just like Klingon, another agglutinating language. One morpheme being surrounded by other morphemes is not what infixing is about. That's just concatenation, pure and simple. Infixing is what happens when a single morpheme gets realized discontiguously, as in semitic stems or Tagalog. For some explosively rich concatenation possibilities see Aymara, which has 24 separate suffix classes for verbs (by one analysis), compared to the nominal 9 suffix classes for Klingon verbs (by the traditional analysis). So you get things like utamancapjjasamachiwa uta-ma-na-ca-pjja-samacha-i-wa "it appears that they are in your house (topic)" Lots of suffixes lined up in a row (uta = house), but not an infix anywhere in sight. Ken Beesley Date: Wed, 5 Aug 92 07:46:11 EST From: j.guy@trl.OZ.AU (Jacques Guy) To: Ken_Beesley.PARC@xerox.com, shoulson@ctr.columbia.edu Subject: Re: Infixing -- in IE too Cc: conlang@buphy.bu.edu Mark Shoulson remarks that "infixing (of the Semitic style) does tend to break IE concepts." Still, there are traces of infixing in Indo-European. Take Greek for instance, where the root "lab" alternates with "lamb", e.g. elabon "they took", lambano "I take". Why, I wasn't thinking of Latin, which has a similar phenomenon, which has even survived into modern Italian e.g. scendere (inf), sceso (past part.). Come to think of it, it is somewhat surprising that conlangs do not resort to infixes. Perhaps infixes are not perceived as such by native speakers, but as alternations, e.g. Greek lab/lamb, Italian scend/sces, just like umlauting or better said vowel alternation, e.g. Latin cap/cep, English goose/geese, foot/feet, etc. I agree that neither Lojban nor Esperanto have anything remotely like infixes, by the way. j.guy@trl.oz.au   Date: Thu, 13 Aug 92 14:19:02 EST From: j.guy@trl.OZ.AU (Jacques Guy) To: conlang@buphy.bu.edu Subject: conlangs? Too complex, all of them! It's only recently, after exposure to Lojban, Guaspi, Interlingua etc, and reminiscing over my dim memories of Esperanto (I was 10 or so at the time) that all these conlangs have struck me as ... (hold onto your seats, this might shock you) ... exceedingly complex. To me, a constructed language should be as simple as possible. That is because I have encountered natural languages that make do very nicely with exceedingly few grammatical categories and the barest syntax. Tolomako (a language of Vanuatu), for instance, makes do, and very well, with a single subordinating conjunction, and two verbal modes (no tenses, no aspects). "Verbal modes" is a bit of a misnomer, because they get applied to nouns, numerals, and adjectives as well. Lolovuevue (Vanuatu again), has a very similar system, but goes one better by completely and very cleanly eliminating the distinction between verbs and adjectives. It considers all as processes, and uses two particles to express that a process is completed, hence a state ("u"), or still going ("mo"). Thus: mo vano "he goes" u vano "he has gone" but: u garea "it is good" *not* "it was good" because: mo garea "it is improving" However simple and easy to learn those languages were, I found it difficult to adapt to them. The restrictions put on you force you to think very differently from English. One example: "Brush your teeth after you've finished eating" becomes a complete, impossible mess if you attempt to translate it like that. Firstly, there is no conjunction, nor preposition for that matter, that means "after". So you rack your brains and come up with: "When you have finished eating, brush your teeth". Not good enough yet. "To finish" is only intransitive. Oh yes, you can make "to cause to finish", but try the result on a native speaker and he'll think you're ready for the lunatic asylum. No, the solution is, literally: "When/if you eat [and the process of eating] is finished, brush your teeth". Speaking Japanese requires you to reorder the words of your sentence. Speaking Tolomako forces you to reorder concepts into chronological and cause-and-effect order. Would make a politician's life rather difficult, I think. Enforced clarity out of outrageous simplicity. What am I getting to? I have come to think that there ought to be at least two stages in designing a constructed language: 1. Make it simple. Done? Now for the second stage: 2. Simplify it. It is far, far, from easy, especially for us, who speak the awful messes known as English, French, German, Spanish, and yes (reach for your flame-throwers) Esperanto, Ido, Interlingua, Lojban, you name them. We have no idea really of how to construct a simple, elegant, powerful language until we are faced with one. Even then, only a dim idea. I find Tolomako, amongst others of its family, wonderfully simple and elegant for a natural language. I still find it far too difficult and messy for an artificial one. Date: Wed, 12 Aug 1992 23:18:01 PDT From: Jonathan Pool Subject: Re: conlangs? Too complex, all of them! To: conlang@buphy.bu.edu Jacques Guy seems to be working in the tradition of philosophical languages, versus Zamenhof et al., who worked for international languages. A language that optimally helps people think interestingly or mystically or logically or some other way won't generally be identical to one that optimally facilitates translingual communication among people whose thinking is still unreformed. But, if this discussion of simplicity is going to remain interesting (at least to me), I'd like to see a formal definition of "simplicity" that permits me to order any two languages according to simplicity. ------- From: helz@ecn.purdue.edu (Randall A Helzerman) Subject: Re: conlangs? Too complex, all of them! To: conlang@buphy.bu.edu (Constructed Language List) Date: Thu, 13 Aug 92 1:45:08 EST X-Mailer: ELM [version 2.3 PL11] > But, if this discussion of simplicity is going to remain >interesting (at least to me), I'd like to see a formal definition of >"simplicity" that permits me to order any two languages according to >simplicity. Please let me be the first to suggest criteria; I wrestle with language complexity daily trying to get computers to understand. :-) 0. The first complexity measure should be where the language fits on Chompsky's heirarchy. Some languages are "harder" to process (in terms of how much computer time it would take to determine whether or not a sentence is grammatical). 1. An obvious criterion is the _size_ of the grammar. Assuming the language is context-free, this would be measured in terms of the number of context-free production rules it takes to specify the grammar. English, for example, is not context free, but you can get reasonable coverage with on the order of 300 or so productions. Hey Lojbab! How many productions are in lojban? The "diagrammed summary" is about 10 pages long. A similar summary of English would be a tome at least a thousand pages long. Anybody else out there in a position to say how many productions are in their conlang of choice? 2. Size of the lexicon is another parameter. Also the complexity of the morphology (how morphemes such as prefixes and suffixes combine to form words) can sometimes be measured in terms of context-free production rules, although I've heard clames that for English this is impossible. Don't really know--Morphology isn't my bag. 3. Syntactic ambiguity of the language is a significant source of complexity. This can be objectivly measured in terms of the average number of parses per sentence. I think the above are reasonable measures, because they are all hurdles one must overcome when actually _learning_ a conlang. Date: Mon, 17 Aug 92 09:30:32 MDT From: mnu@inel.gov (Rick Morneau) Subject: Transitivity and Affricates Howdy conlangers! At the risk of sounding somewhat "Darwinian", as Ken Beesley puts it, I will very hurriedly try to add some data to the discussion on transitivity. It's certainly not practical to do a real language count, since data for most languages spoken on this planet is simply not available. However, I have some familiarity with the transitivity aspects of several of the more widely spoken languages. By the way, all of what follows is from my less-than-perfect memory. If I had more time, I'd do a better job, but at the moment I don't have that luxury. If I'm wrong, correct me - don't shoot me. :-) Basically, if we limit ourselves to "major" languages, then we would conclude that transitivity is almost always marked. In other words, for the large majority of the worlds major languages, a verb cannot be used both transitively and intransitively without some morphological or lexical indication of the change in usage. The only language I know of that goes against this trend in a regular and productive manner is Chinese. For all others (including English), such freedom is either totally denied, or is irregular and/or idiomatic. English falls into the latter category, as I will illustrate later. The most common way of marking transitivity is similar to the way Esperanto does it; i.e., affixes convert from transitive to intransitive or vice-versa. Turkish, Quechua, and many, many others fall into this category. (Ken adds Mongolian, which, I believe, is in the same family as Turkish.). Some languages, such as Swahili, perform the conversion only in one direction; i.e., they have verb roots that are inherently transitive and add an affix to convert them to intransitive verbs. I'm pretty sure, but not certain, that Japanese and Korean are also of this type. (Sorry, but I'm doing this from memory). Also, if I remember correctly, the Dravidian languages such as Tamil, Telugu and Kannada go in at least one direction, but perhaps both. (If Dorai Sitaram is still on this list, perhaps he can comment on this. I'm also curious what Hindi does.) Other languages, such as Arabic and Indonesian (and how about Indonesian's close cousin, Tagalog, Ken?), have basic roots that undergo morphological derivation to create transitive and intransitive verb forms. In fact, Arabic makes more distinctions than any other language I am aware of (transitive/causative, intransitive/reflexive, causative reflexive and one other that I've forgotten), although these derivations may no longer be as productive as we would like for a conlang. Romance languages such as French and Italian use reflexive clitics, and Russian (and other Slavic languages???) use reflexive affixes. German (and also Dutch???) allows some freedom, but not as much as English - it often uses a reflexive construction, or morphologically derives a new verb with a prefix. Can anyone comment on the Scandinavian languages? Probably ALL languages have irregular forms, such as English "kill/die" and "drop/fall". English is not nearly as productive as you may think. In fact, for many verbs, it is very nearly idiomatic. Consider the following examples: He broke the window. = transitive The window broke. = intransitive He smashed the window. = transitive *The window smashed. = intransitive (ungrammatical) He built the doghouse. = transitive *The doghouse built. = intransitive (ungrammatical) He painted the doghouse. = transitive *The doghouse painted. = intransitive (ungrammatical) She flew the plane. = transitive The plane flew. = intransitive She drove the car. = transitive *The car drove. = intransitive (ungrammatical without adjuncts) He climbed the mountain. = transitive *The mountain climbed. = intransitive (ungrammatical) The tree fell. = intransitive *He fell the tree = transitive (ungrammatical) In other words, it works for some but not for others, even when it makes perfectly good sense semantically. By the way, I believe that ALL of the above would be grammatical in their corresponding Chinese forms. (I'm referring specifically to Mandarin Chinese, but I am unjustifiably assuming that other Chinese languages may behave similarly. As I said earlier, if I'm wrong, correct me - don't shoot me. :-) However, I still agree with Rick Harrison and Ken Beesley that transitivity in Esperanto is essentially irregular, and that Esperanto has seriously failed in its goal of regularity for other aspects of the language as well. The only feature of Esperanto that is regular is its inflectional morphology. All other components of the language are as irregular as the languages that it borrowed from. Moving to another topic recently discussed, "ts" can be pronounced as a single phoneme or as two separate phonemes. If it is pronounced as a single phoneme, then it is called an _affricate_. Most English speakers will pronounce "ts" as an affricate in "tsetse fly", but will pronounce it as two distinct phonemes in "but see". A better example of affrication for English speakers is the sound of "ch" in "church". This is a true affricate. Furthermore, it is articulated ALMOST as if it were a sequence of the two phonemes "t" in "tan" and "sh" in "ship". Because of this, the IPA symbols for affricates appear to be digraphs, but are in fact single characters. Note, though, that there is a difference! The "ch" in "church" is NOT the same as the "t"+"sh" of "hot shot". The sequence "ks" cannot be a true affricate because its components are articulated in positions that are too far apart. ("k" is velar and "s" is alveolar.) Note that "t" and "s" are both alveolar, and "sh" is palato-alveolar which is very close to "t". Thus, a digraph is appropriate for "ks", while a single character is appropriate for a true affricate. Regards for now, gotta run, my apologies for obvious sloppiness, Rick -- *=*= Disclaimer: The INEL does not speak for me and vice versa =*=* = Rick Morneau Idaho National Engineering Laboratory = * mnu@inel.gov Idaho Falls, Idaho 83415, USA * =*=*=*=*=*=*=*=*=*=*= NeXT Mail accepted here! *=*=*=*=*=*=*=*=*=*= Date: Wed, 19 Aug 92 13:44:00 EST From: j.guy@trl.OZ.AU (Jacques Guy) To: conlang@buphy.bu.edu Subject: "drop" (was: Transitivity and Affricates) Jim Cowan on the subject: ----- start quote ----- "Drop" is a puzzle that was drawn to my attention when I needed the word recently ("the monkey dropped a coconut"). An over-hasty use of the dictionary gave me "dirgo", which unfortunately means "a drop (of liquid)"! However, "cause to fall" is not the same as "drop". When I drop something, I do not cause it to fall in any physical sense; the cause of its falling is the mutual gravitational attraction between the object and the Earth. I don't know what the right answer is yet. ----- end quote ------- To release, I'd say. French has "laisser tomber" (let fall) for "drop", distinct from "faire tomber" (cause to fall), which translates, amongst others, "to trip (someone)". Sakao has "ptjo/" which means "let go, release", and also "throw". "Throw????" will you wonder. Think a bit. Throwing implies releasing. A difference of viewpoint, that's all. Just as legitimate as "throw", for which they have no general word, for they think of the target, so that the word for "throw" is not our "throw" but "hit with a missile". Date: Fri, 21 Aug 92 19:01:53 +0200 From: maxwell@ltb.bso.nl Baudouin de Courtenay (1845-1929) is often considered one of the founders of modern linguistics. As a linguist, I guess he's in the same league with Jespersen, even though he is less well-known in the west, since he lived in Eastern Europe. Even though he didn't develop or support any specific conlang himself, he took an interest in them and provided the following ideas in a speech given at a meeting (called in honor of Zamenhof, who had died some time the previous year) of the esperantist society "Espero", in Petersburg on 14 April 1918. The following are excerpts from this speech, provided in *Impeto'89* a publication of the then-Soviet section of "Mondpaca Esperantista Movado"(World peace Esperanto movement). I hope that not too much has been lost in translation (Russian--Esperanto--English). My comments in [].-- About 40-50 years ago the idea of creating an artificial language was considered ridiculous. It was thought that humans cannot consciously intervene in the evolution of language and change this evolution. [this view is still found today]. The dominant opinion among scholars was that, because language is something "natural", human consciousness cannot reshape it. Later, unconsciously, like all great ideas, the idea was born that because language is an instrument, humans not only may, but even have a moral duty to perfect this instrument. This idea was the forerunner of the creation of an international language and in work on this task two viewpoints are reflected. The first is provided by Schleyer, creator of Volapuk, one of the first successful artificial language projects. But he strove to have Volapuk as world language replace all living languages. [It becomes clear in the following passage that BdC means that Schleyer wanted to make Volapuk the only language in the world, not just the only world language. Does anybody know more about this?] This is certainly an unrealizable utopia. Such dreamers supposed that with the introuction of such a unique language a united humanity would also be created, and then any sources of strife among humans would disappear at the same time and eternal human harmony would reign. Unfortunately, he kept forgetting that besides the differences between peoples and governments, there exist within single populations social and class distinctions. The second view of artificial langauges is more modest. According to it, the role of an international language is not to drive back other languages, but to be a helping language, conserving the language of each individual. And precisely this task was the one which Dr. Zamenhof established and solved. ........ When this task [of perfecting an international auxiliary language] is recognized by those specialists whose duty in this respect is greater than that of all others, that is, those who specialize in research on language and languages, then from the international auxiliary languages, among which Esperanto occupies the first position, a final perfect form will be worked out. Dan Subject: IAL desiderata From: jwt!bbs-hrick@peora.sdc.ccur.com (Rick Harrison) Date: Sat, 29 Aug 92 22:03:18 EDT Here's the first draft of a proposed list of desiderata for an IAL. If we can come up with a list of objective criteria on which most reasonable people can agree, we can use it to evaluate existing IAL proposals and hopefully to influence future IAL designers. I invite your comments; do you see any factual errors or faulty reasoning here? Note: no existing IAL proposal that I know of meets all these criteria. ----- Guidelines for the design of an international auxiliary language 1st draft, 29 August 1992 by Rick Harrison An international auxiliary language for world-wide use must be designed according to reasonable linguistic criteria rather than guesswork or an individual's whims. This document is an attempt to list such criteria. We begin with the assertion that the international auxiliary language (IAL) must be relatively easy for most humans to learn and use. Three factors that make a language relatively easy to learn are simplicity, regularity, and familiarity. (Simplicity means, for example, that the rules of grammar should be few. Regularity means, for example, that there should be no allomorphy. Familiarity means, for example, that the language should not use phonemes which are unfamiliar to most of the world's people.) All the guidelines listed below are based on these principles. 1. Orthography 1.1 writing system The four most widely distributed writing systems are: the Latin/Roman alphabet, the Cyrillic alphabet, the Arabic alphabet, and the system of logograms common to Chinese, Japanese and Korean. The IAL should use one of these systems without adding any unique characters or symbols from other alphabets which could cause typographical difficulties. The Latin alphabet is the most widely distributed and is probably preferable, but valid arguments can be made for the other options. Simplicity and regularity are desirable in the orthography of an IAL. Therefore, in alphabetic writing systems, each letter should represent only one phoneme, and each phoneme should be represented by only one letter. Digraphs are unacceptable. 2. Phonology 2.1 phoneme selection The IAL should only use phonemes which most people in the world already know or can easily learn to pronounce. In other words, the choice of phonemes must be based on an objective survey of the sounds used in the world's major languages, not on the personal preferences of an individual language inventor. A demonstration of how this selection process should be performed follows. The set of vowels should be limited to the "cardinal" vowels a, e, i, o, u. Other vowels are relatively rare in the world's languages. Diphthongs are unacceptable for two reasons: 1) they have to be represented by digraphs or special characters, and 2) speakers of some languages tend to pronounce certain vowels as diphthongs (for example, many English speakers cannot produce a pure "o" and instead tend to say "ou"), therefore the use of diphthongs as elements of an IAL could make correct pronunciation unnecessarily difficult. Consonants should also be limited to those which can be easily produced and distinguished by the majority of human beings. The consonants [k] [m] [s] [t] occur in 100 percent of the world's 25 major languages (Morneau 1991). If we allow the letter "p" to be pronounced as [p] or [b], it also represents phonemes present in 100% of these languages. (It is desirable to allow this sort of allophonic variation to make pronunciation easy for as many of the world's people as possible.) The consonants [l] [n] and the semi-vowel [j] exist in 96% of the 25 languages surveyed. The IAL should allow the letter "r" to be pronounced as an alveolar trill, a uvular trill, or a post-alveolar or retroflex approximant; these "rhotic" sounds exist in 92% of the 25 languages. Allow "h" to be pronounced as [h] or [x], and allow "f" to be pronounced as [f] or [v]; these phonemes are used in 88% of the surveyed languages. The "sh" consonant heard in "ship," the semi-vowel [w], and [z] are also widely distributed enough that they might be accetable in an IAL. In Mandarin Chinese, spoken by more people than any other language, g = [k] and k = [k'], d = [t] and t = [t']. This is quite different from the voiced-unvoiced distinction made between these letters in European languages. For this reason, "d" and "g" should probably be omitted from the IAL. (Alternatively, "d" and "g" could be included if "t" and "k" are omitted.) In summary, the alphabet of an IAL which is designed to be maximally easy for most humans to pronounce should be something like this: a e f h i j k l m n o p r s t u or a e f h i k l m n o p r s t u y or a c e f h i j k l m n o p r s t u w z or a c e f h i k l m n o p r s t u y w z ...where "c" represents the "sh" phoneme; [j] might be represented by "j" or "y." 2.2 consonant clusters For many of the world's people, consonant clusters such as /str/, /spl/, and /kt/ are difficult or impossible to pronounce. Therefore the structure of syllables in the IAL must be controlled. Consonant clusters should be totally eliminated, or limited to those which are universally easy to pronounce. If the latter course is chosen, the following guideline should be used: A syllable may begin with a vowel, or with any single consonant, or with a consonant followed by [j] or [w]. A syllable may end in a vowel or a continuant consonant. (Continuants are nasals, liquids, and fricatives.) 2.3 tones Some languages use tones to distinguish syllables which are otherwise identical; for example, in Chinese _ren_ means "a person" when pronounced with a rising tone, but means "to recognize / identify" when it has a falling tone. This sort of tonal distinction occurs in a minority of natural languages and makes those languages more difficult for others to learn. Therefore, the IAL should not use tones in this manner. to be continued... Date: Sat, 29 Aug 92 23:42:15 -0400 From: lojbab@grebyn.com (Logical Language Group) Subject: Re: IAL desiderata Why does allomorphy violate regularity as a principle? If all allomorphs mean the same thing, then allomorphs that fit a regular system ARE regular. You might argue that simplicity is violated more easily than regularity, but even there a case can be made against it. Specifically, by the nature of language, and in particular a conlang to be used internationally, you will be dealing with many different accents of the different native language speakers. A conlang will have to have considerable tolerance therefore for ALLOPHONY. Now if speech is considered to be the basic form of language, than a phonetic representation of even an optimal conlang will therefore show ALLOMORPHY in that different allophones might be represented differently to convey what is actually being said. Now I don;t advocate having different spellings for different allophonic pronunciations; I merely claim that if you have allophony, than at some level you in effect have allomorphy - people of other language/accent backgrounds are going to hear what you say as being different from the way they would say it. Sometimes these differences will extend to the point that they will be heard as different words althogehter because of the different phonemic mappings of the different languages. A good conlang can define away a lot of these problems, but you can always find some speakers in the world who will have a trouble with any phoneme set you come up with. Since we CAN understand people who speak with different accents, whether non- natives speaking English, or Chines and Europeans speaking Esperanto, I claim that the human language skill has some amount of ability to automatically handle allomorphs. I don't think the human mind clearly processes at the individual sound level, but rather on the word level. We are capable in human speech of filtering out noise syllables (um's and er's and ya know's), as well as the difference between cannot and can't, and even the British vs. American pronunciations of garage, and doing so totally at the subconscious level. Now of course I'm a bit prejusdiced in this in that Lojban has allomorphs, all of which mean the same thing. But of course the likelihood as we've found is that there are only two allomorpjhs that seem to get used for any given lujvo compound (the things that have allomorphs). One is some optimal short form that gets adopted through heavy usage, and is usually one of the shortest possible forms available, and the other is the minimally reduced form used in noisy environments or when talking to beginners in the language that don't know the reduced forms. The virtue of having THESE two allomorphs is, I beleive, indisputable. One makes the language easy to use with minimal learning; the other makes the language agreeable concise when you know it well. (Given that one of the few INTERNATIONAL LANGUAGES that IS universally accepted is the ALPHA/BRAVO/CHARLIE system of spelling out words in noisy environements, to condemn us for this is to condemn an accepted practice used far more than even Esperanto. Anfd the dichotomy matches Lojban's quite closely. Now Lojban does have more allomorphs than two - these are primarily used when words are first being created and no single short form has been agreed upon by consensus. This is of course a problem that will go away in time, t though perhaps not - it makes poetry much easier, something that your optimal conlang will not so easily support due to its sparseness. But they also allow for the human need to adapt in the face of redundancy problems, difficult sound combinations, etc. Any rigid and firm single-form system will almost certainly in some cases generate combinations that are unaesthetic or even unpronouncible, OR which can be confused with some similar sounding word that occurs in similar contexts. In these situations, which will occur in any conlang, having an allomorphic form to switch to may keep the language able to be USED. The theoretical criteria you offer are merely ideals, and must give way before the needs of human speakers if your language is truly to be "universal". (Indeed, come to think of it, I suspect that for aesthetic reasons, all of your criteria might be questioned by people with different backgrounds and philosophies, not just this one that I have raised. For example, I have argued that high recognition of words can be a detriment to learning because it causes semantics transfer, and no well desighned conalng is likely to have the exact same semantics for such words as the natlang which is being confused with.) Enough for now. I suspect that you will not be able to get people to agree on any set of criteria based on that last line of reasoning. I certainly don't agree. lojbab Subject: Re: IAL desiderata From: Jim Gillogly Date: Sat, 29 Aug 92 21:04:06 PDT As long as you're speccing a whole new project... For a language intended to be easy for virtually everybody to pronounce, I think you really shouldn't keep 'r' and 'l' as separate consonants. The distinction is not at all familiar to Japanese, and, although that's only one language group, it's an important one. I'd say keep one liquid, but allow either pronunciation, or, as in Japanese, any pronunciation in between them. Jim Subject: IAL desiderata, part 2 of 2 From: jwt!bbs-hrick@peora.sdc.ccur.com (Rick Harrison) Date: Sun, 30 Aug 92 04:03:16 EDT Guidelines for the design of an international auxiliary language part 2 3. Morphology and Word Allocation There are many reasonable ways to design the structure of morphemes in a planned language. Should a language use special endings, or the arrangement of consonants and vowels, to indicate whether a given word is a verb or noun? Should the creation of compound words be allowed? Should bound affixes be different in some obvious way from free roots? Reasonable persons might disagree on these matters. However, the following guidelines seem desirable. 3.1 general Morphemes should be constructed or allocated in a manner which ensures that no single morpheme looks and sounds identical to a compound of two or more morphemes. 3.2 compounding If the IAL permits the creation of compound words, the meaning of any compound must be deducible, on the basis of logic and pragmatics, from the meanings of the individual formatives. Compounds such as "cranberry," in which one formative has no independent meaning, should not exist. (Maxwell 1989) In a language of CV and CVCV roots, a CVCVCV compound could possibly be analyzed as: CV + CV + CV _or_ CV + CVCV _or_ CVCV + CV. This type of morphological ambiguity is unacceptable. The morphology must be designed so that there is only one way to dissect a compound word into its constituent parts. 3.3 allomorphy Allomorphy refers to one morpheme having two or more variant forms; for example, the plural suffix in English is -s in "books," -es in "boxes," and -en in "oxen." Allomorphy is an unnecessary irregular- ity which increases the difficulty of memorizing a language's vocabulary; therefore allomorphy is unacceptable in an IAL. 3.4 synonymy The vocabulary of root-words should not contain any synonyms. 3.5 polysemy Each morpheme should represent a single semanteme (unit of meaning) or a series of closely related semantemes. No individual morpheme should represent several disparate meanings. 3.6 brevity In natural languages, the most frequently used morphemes tend to be the briefest morphemes; seldom-used morphemes tend to be longer. An IAL should exhibit the same tendency. 4. Grammar 4.1 universality The grammar of the IAL should be a streamlined distillation of those features which are nearly universal in the world's major languages. A constructed language with a grammar which is totally different from the grammar of any natural language might be a useful experimental tool but is not a worthy candidate for the role of IAL. 4.2 syntax Linguists have attempted to classify natural languages based on the arrangement of subject (S), verb (V) and object (O). There is some debate as to whether this system can accurately describe all languages, but it is sufficient for our purposes. "There are six possible orderings: VSO, SVO, SOV, VOS, OVS, and OSV. It turns out that a very large majority of the world's languages fit within the first three categories; i.e., where the subject comes before the object." (Morneau 1992) Therefore the IAL should either be designed to allow all three of these sequences (VSO, SVO, SOV) to be used, or should adopt SVO as its sequence. SVO is the predominant word order in the world's major languages, and is notably common in languages which are used as interlinguas between different language groups (e.g. English, French, Swahili, Indonesian). A language with a relatively strict SVO ordering is efficient because it does not require the use of inflections or marker words to distinguish subject from object. For these reasons, SVO is probably best for an IAL. 4.3 gender One factor which complicates the learning of languages such as French and German is the need to memorize the gender of every noun. The IAL should either treat all non-living things as neuter or should entirely discard grammatical reflection of gender. 4.4 transitivity In some constructed languages, it is necessary to memorize whether a verb is inherently transitive or intransitive, and affixes are used to convert transitive verbs to intransitive and vice versa. Having to memorize a verb's transitivity is an unnecessary burden, just as having to learn the gender of a noun would be. Therefore the IAL should either form transitive and intransitive verbs in some completely regular and predictable manner, or should permit unvarying verb morphemes to have variable transitivity (as is the case with most English and Chinese verbs). 5. Bibliography Maxwell, Dan: Principles for constructing planned languages. in: _Interlinguistics_ edited by Klaus Schubert; New York & Berlin: Mouton de Gruyter, 1989 (348 p.) Morneau, Rick: Designing an artificial language, part 1. September, 1991. Distributed by electronic mail. Morneau, Rick: Designing an artificial language -- Syntax. August, 1992. Distributed by electronic mail. -fin- Date: Sun, 30 Aug 1992 09:57:02 -0700 From: D Anton Sherwood Subject: Antonese Hello from a new name on the list. I've had the idea of an experiment in accelerated historical linguistics: A journal is published in a constructed language. From time to time the editor decrees a sound shift, and runs a string-transform on articles submitted for the current issue. Words not used in the current issue are "archaic" and therefore not affected by that change (but may be affected by subsequent changes). Contributors are encouraged to invent words -- by using archaism (borrowing from old versions), compounding, ablaut and the like, or (if all else fails) dice tables. Direct borrowing is forbidden, but calques are tolerated. From time to time, of course, the editor publishes the lexicon; ideally it is available on line. --- Here's a syllabary for the language I haven't designed yet, which I call Antonese. Eight initials: 0 p t s r k h m Thirteen kernels: i e a o u ja jo ju aj aw wi we wa Three terminals: 0 n + /n/ can be considered as a consonant (in which case it assimilates to a following stop) or as nasalizing the vowel. /+/ adds a generic mora, either by lengthening the vowel or by geminating the following consonant. My goal in constructing this syllabary was maximally distinct phonemes (few humans would confuse any of these, I think) without making the syllabary too impoverished. This system is very tolerant of "foreign accents": it doesn't matter if you voice the stops, pronounce /r/ as /l/, or round the wrong vowels. --- Anton Sherwood == dasher@well.sf.ca.us Date: Mon, 31 Aug 92 10:47:22 EST From: j.guy@trl.OZ.AU (Jacques Guy) Subject: Re: IAL desiderata Random thoughts. Everything desirable can be undesirable for others. Take phonology. Limiting the phonological inventory to what is as much as possible acceptable to all leads to few phonemes. Limiting consonant clusters leads to a (C)V language. So far so good, or so does it seem. Well, let's take such a real language: Rotokas (New Guinea). Five vowels: a e i o u, six consonants: g k p r t v. No consonant clusters, no closed syllables. To me it is a frightful tongue-twister. ouokivuia ragai ibu iare avaraepa ogoevira ikauoro eakepa viapau rera kaakau taparevora voari When I try to utter those sentences (they are *real* Rotokas) my tongue trips all over itself like the proverbial millipede who was asked: which leg do you put forward first? I need consonant clusters to be comfortable, but that is unfair to those whom consonant clusters make uncomfortable, and I am not sure that a compromise somewhere in-between is not taking the worst of two bad worlds. Perhaps a language that can be read out in different ways? Latin was such when it was an international language; each nationality would pronounce Latin as if its spelling followed the rules of their own languages. Not too desirable a state of affairs, but... image a IAL with three vowels, a e o, and consonant clusters. You're left with two good, serviceable, vowels with which its consonant-shy speakers can break up clusters if they so wish. Me, it's vowel clusters that make me trip, give me a spare consonant please to break up those clusters -- not [h], I run out of breath easily; not [x] I'd end up with a sore throat... a difficult customer. I *told* they were random thoughts, raw, unfollowed through... Subject: re: IAL desiderata From: jwt!bbs-hrick@peora.sdc.ccur.com (Rick Harrison) Date: Mon, 31 Aug 92 00:04:15 EDT trl.OZ.AU!j.guy (Jacques Guy) writes: > Limiting consonant clusters leads to a (C)V language. So far > so good, or so does it seem. Well, let's take such a real > language: Rotokas (New Guinea). Five vowels: a e i o u, > six consonants: g k p r t v. No consonant clusters, no > closed syllables. To me it is a frightful tongue-twister. > > ouokivuia ragai ibu iare avaraepa ogoevira ikauoro I agree that such a sentence is hard to pronounce, and I think a lot of the world's people would have trouble with it. However, you seem to have ignored the criteria that I actually proposed: >Consonant clusters should be totally eliminated, or limited to >those which are universally easy to pronounce. If the latter >course is chosen, the following guideline should be used: A >syllable may begin with a vowel, or with any single consonant, >or with a consonant followed by [j] or [w]. A syllable may end >in a vowel or a continuant consonant. (Continuants are nasals, >liquids, and fricatives.) While a lang designer working within these constraints _could_ come up with something like Rotokas, he would more likely (in view of the need for more brief morphemes and other criteria) come up with a language that has sentences like this: a palun wa kwito yunza swi ramfe. There is evidence that this type of structure closely approaches the built-in, instinctive linguistic tendencies of the human brain -- in other words it approximates what is 'burned into the ROM chips' of our computers. The particular phonemes and combinations which are learned relatively late by children also tend to be relatively rare in the world's languages, tend to be among the first sounds to disappear in aphasics, and tend to be absent in glossolalia: ''While the proportion of open or closed syllables does vary considerably between languages, there are no languages without open syllables. On the other hand many languages, particularly in the Pacific and the Indian Ocean, do not allow consonant clusters, causing native speakers to intercalate vowels in borrowed words to restore CV alternations. ''Now the tendency to have open syllables, and the more or less total exclusion of consonant clusters, even in initial position, are striking characteristics of glossolalic output, whatever the mother tongue of the speaker happens to be. They are similarly dominant in child language.'' - Marina Yaguello, _Lunatic Lovers of Language_. -hrick Date: Mon, 31 Aug 92 09:08:04 MDT From: mnu@inel.gov (Rick Morneau) Subject: Re: IAL desiderata Howdy conlangers! Kudos to Rick Harrison for his attempt to bring order out of chaos! I certainly agree with most of the points he made, and hope that he doesn't mind if I bring up a few minor disagreements and nitpicks. First of all, a nasal is not a continuant. This is the same mistake that I made in one of my earlier essays, and someone kindly corrected me. Thus, your sentence: > > A syllable may end in a vowel or a continuant consonant. > (Continuants are nasals, liquids, and fricatives.) > should be corrected to read: "A syllable may end in a vowel, a nasal or a continuant consonant. (Continuants are liquids and fricatives.)" I disagree with your judgement against diphthongs. I might agree to avoid diphthongs articulated at two very close vowel positions, as in your example "ou". However, other diphthongs are easy to pronounce, highly distinctive, extremely efficient, and very common. Thus, I would definitely include diphthongs such as /wa/, /aw/, /ay/, /ya/, /oy/, /yo/, /we/, /wi/, and perhaps /ew/ and /iw/ (where /w/ and /y/ are semivowels). I would not discard /d/ and /g/ just because of the Chinese and Koreans. Even though these sounds do not exist as phonemes in those languages, they DO exist as allophones, and both Chinese and Koreans seem to have little difficulty learning to distinguish them from their unvoiced counterparts. The only problem you may run into is in teaching Koreans /z/, since their language does not voice ANY of its fricatives, even allophonically. Keep in mind that, if you give every language veto power, you'll end up with next to nothing. The essay doesn't state it explicitly, but you seem to be saying that polysemy is bad. I don't see how you can avoid polysemy, as long as you don't stretch meanings to the breaking point (say, by using the same word with different argument structures). The only alternative to polysemy will be a vocabulary that is so large that it is unlearnable, since you'll need a different word for every slight shade of meaning. Besides, humans use polysemy automatically, and I doubt if you could train people out of the habit. Also, by barring polysemy, you implicitly bar metaphor, since polysemy covers the mid-ground between literal meaning and metaphor. I can't imagine any language that disallows metaphor. One point I feel you should have made (or should have stated more forcefully) is that an IAL must be as neutral as possible. An IAL that is heavily based on a few closely related natural languages (such as Esperanto, Ido, Glossa, Interlingua, Novial, etc.), will be MUCH easier to learn for some than for others. My personal feeling is that a credible IAL must be just as easy to learn for a Chinese or an Indonesian as it is for an Arab or an Italian. Finally, I'd like to comment a little on the discussion of allomorphy. As best as I can discover, allomorphy exists in natural languages either because of historical accident, phonological rules or cultural requirements. For example, the English plural morphemes "s" and "es" and Hungarian and Turkish vowel harmony all show what appear to be "irregularities", but are actually based on phonological requirements. Some inflection in Arabic, French and Russian show irregularities that exist for both phonological and historical reasons. (English "en" in "oxen" is, of course, a historical hangover). And culturally, the use of different morphemes with essentially the same meaning are used to indicate register (i.e., relative status between speaker and listener), as in Japanese and Cambodian. We don't see "true" or "unbiased" allomorphy simply because languages evolve, and if two morphemes initially have the same meaning, they will eventually drift apart and take on different meanings. This is a well-know linguistic phenomenon, and no language can avoid it. Thus, I agree totally with Rick Harrison that allomorphy is definitely a BAD THING, and should be avoided in any IAL. However, I'm not sure that this restraint should apply to the loglans. Their main purpose, after all, is to test the Sapir-Whorf Hypothesis. In order to do this, they've designed a language that seems to go against many linguistic universals. And introducing a new form of intentional allomorphy is just another example of the differences between loglans and natural languages. Thus, allomorphy may be OK for a loglan even though it is undesirable for an IAL. Regards, Rick -- *=*= Disclaimer: The INEL does not speak for me and vice versa =*=* = Rick Morneau Idaho National Engineering Laboratory = * mnu@inel.gov Idaho Falls, Idaho 83415, USA * =*=*=*=*=*=*=*=*=*=*= NeXT Mail accepted here! *=*=*=*=*=*=*=*=*=*= Date: Mon, 31 Aug 92 09:33:51 MDT From: mnu@inel.gov (Rick Morneau) Subject: Re: IAL desiderata Jacques Guy writes: > > I need consonant clusters to be comfortable, but that is > unfair to those whom consonant clusters make uncomfortable, > and I am not sure that a compromise somewhere in-between is > not taking the worst of two bad worlds. > I believe that a compromise IS possible. First, do not allow consonant clusters WITHIN a syllable - a cluster should only be allowed to straddle a syllable boundary. Second, syllables should be open, or should only be closed with nasals or continuants. Third, if a closing consonant is not a nasal, then it should be voiced/unvoiced the same as the first consonant of the next syllable. If it IS a nasal, then it should be homorganic with the first consonant of the next syllable. Thus, syllable structure would be: syllable ::= (C)V(V)(X) C = consonant V = vowel X = L or N or F L = liquid N = /n/ before /z/, /s/, /d/, /t/, /j in judge/, /ch in church/ /m/ before /v/, /f/, /b/, /p/ /ng in sing/ before /g/ or /k/ F = /s/, /f/ or /sh in ship/ if the following C is unvoiced, otherwise /z/, /v/ or /s in vision/, respectively. For those of you who are unfamiliar with basic phonology, the above may appear somewhat daunting. If so, play with the combinations for a while, and you'll see that they make very good sense. Basically, they all reflect very simple rules of consonant harmony, most of which exist in English and which speakers of English use automatically. Regards, Rick -- *=*= Disclaimer: The INEL does not speak for me and vice versa =*=* = Rick Morneau Idaho National Engineering Laboratory = * mnu@inel.gov Idaho Falls, Idaho 83415, USA * =*=*=*=*=*=*=*=*=*=*= NeXT Mail accepted here! *=*=*=*=*=*=*=*=*=*= From: "Edmund Grimley-Evans" Date: Mon, 31 Aug 92 21:13:06 +0200 To: conlang@buphy.bu.edu Subject: IAL desiderata > Some languages use tones to distinguish syllables which are otherwise > identical; for example, in Chinese _ren_ means "a person" when > pronounced with a rising tone, but means "to recognize / identify" when > it has a falling tone. This sort of tonal distinction occurs in a > minority of natural languages and makes those languages more > difficult for others to learn. Therefore, the IAL should not use > tones in this manner. This is untrue. (1) The *majority* of the world's languages use tones. (2) Chinese children learn to pronounce the tones before they learn the consonants and vowels properly. It is *not* unconditionally true that tones "make ... languages difficult ... to learn". It may be more difficult to learn tones from a book, but if you're not learning from a book it's easier to learn tones than to learn just about any other sound. Ask any blind person who has learnt Chinese as a foreign language if you don't believe me! Date: Mon, 31 Aug 92 14:27:33 MDT From: mnu@inel.gov (Rick Morneau) To: conlang@buphy.bu.edu Subject: Re: IAL desiderata Howdy conlangers! A discussion of tone languages on conlang list? Who would have thought it possible! :-) "Edmund Grimley-Evans" writes: > > (1) The *majority* of the world's languages use tones. > Yes, indeed. But NOT in the way that Rick Harrison was talking about. Using your over-broad definition, even English would be considered a tone language, since all languages use tone (i.e., pitch and pitch contours) for some purposes. However, only a small minority of the world's major languages use tone to make PHONEMIC distinctions. This fraction does get larger if you include more obscure languages (if memory serves, most tonal languages seem to be in the Niger-Khordofanian family of Africa, especially the Bantu languages). Also, keep in mind that many of these (especially African languages) have only one or two non-neutral tones, and use them for syntactic, rather than lexical distinctions. But even if you were targeting your conlang at existing tone-language speakers, you would still face serious problems, since different tone-languages use tones differently, and knowing one system does not necessarily make it easier to master a different system. > > It is *not* unconditionally true that tones "make ... languages > difficult ... to learn". > I disagree with this even if you use your over-broad definition of a tone language. And when you narrow down the definition to include only conventional "tone languages", you're going to have an even more difficult time convincing anyone. I've studied both Mandarin Chinese and Vietnamese. The tonal distinctions were, by far, the most difficult things that I've ever encountered in my language studies. And even after I thought I had mastered the tones of specific words, I constantly found myself forgetting them, and slipping into the wrong tones. It ain't easy, buddy! Actually, this difficulty shouldn't be surprising, since the use of tones to make phonemic distinctions is very foreign to the way I normally speak. And language-teachers have long known that learning how to make distinctions that you are not used to making is one of the most difficult aspects of language-learning. Rick Harrison was correct - stay away from tones if you want to make your IAL as easy to learn as possible! Regards, Rick -- *=*= Disclaimer: The INEL does not speak for me and vice versa =*=* = Rick Morneau Idaho National Engineering Laboratory = * mnu@inel.gov Idaho Falls, Idaho 83415, USA * =*=*=*=*=*=*=*=*=*=*= NeXT Mail accepted here! *=*=*=*=*=*=*=*=*=*= From: cbmvax!snark.thyrsus.com!cowan@uunet.UU.NET (John Cowan) Subject: In defense of grammatical gender Date: Mon, 31 Aug 92 12:08:47 EDT Rick Harrison writes: > One factor which complicates the learning of languages such as French > and German is the need to memorize the gender of every noun. The IAL > should either treat all non-living things as neuter or should entirely > discard grammatical reflection of gender. Gender has its uses. A good example (due to Quine): He removed the manuscript from the briefcase and cast it into the sea. The English doesn't make clear what went into the sea. A French version, however, is unmistakable: Il retira le manuscrit de la serviette et le (la) jeta dans le mer. "le" means "le manuscrit" and "la" means "la serviette", no ambiguity. Now of course it is annoying to have to learn the gender of nouns painfully one by one, but Rick chooses the hard cases German and French to make his point! In Spanish, Italian, and Russian, morphology pretty nearly determines gender, and the above advantage is still preserved. Lojban (toot, toot) has ten "genderless" pronouns corresponding to "he/she/it", which can be used to talk about up to ten things at a time, but which must be explicitly assigned to their referents. However, it is also correct to use letters of the alphabet as pronouns, and then they are assumed to refer to the most recent referent beginning with that letter. It would therefore be accurate to say that Lojban has 17 grammatical genders: the 'b' gender, the 'c' gender, the 'd' gender, ..., each with its gendered pronoun. -- John Cowan cowan@snark.thyrsus.com ...!uunet!cbmvax!snark!cowan e'osai ko sarji la lojban. Date: Tue, 1 Sep 92 10:03:17 EST From: j.guy@trl.OZ.AU (Jacques Guy) To: conlang@buphy.bu.edu Subject: Re: IAL desiderata: tones et al. The variety of opinions about the desirability of tones strikes me. I am for all purposes musically tone-deaf -- a misnomer, but never mind. I can sing only off-key, out-of-tune, the screech of an owl is sweeter music than my singing. Yet I had relatively little trouble mastering the four tones of Mandarin. Why? Thinking back, I'd chalk it up to my mother tongue, French, in which intonation plays a much smaller part than in English. E.g.: I did it = Je l'ai fait; *I* did it = C'est moi qui l'ai fait, or: Celui qui l'a fait, c'est moi, etc., etc. So that, for instance, the third tone did not so strongly instinctively convey the interrogative sense it does in English. That said, it is a pity that there is much, very probably valid, reluctance towards tones. Tones are good for speech recognition, I'm told by a colleague here who's into it (and consonants are bad, especially nasals and voiceless stops), and speech recognition is fashionable. Now for "et al", which is about clusters of vowels, or consonants. {CV(V)} I can cope with, it's when it becomes {(C)V(V)} that my tongue and my ear lose all sense of direction. Of consonant clusters, I somehow feel that a stop preceded by a homorganic nasal is acceptable to about everyone, provided the combination is flanked by vowels on both sides, e.g. ambo, anta, etc. Next would be perhaps stop+liquid with a vowel following: tra or tla, bla or bra. And close, very close behind, stop+sibilant: tsa, dza, psa of pfa. Finally, an important piece of trivia. In 1974 I was in Lolovuevue (Lepers' Island a.k.a. Omba, Aoba, or Oba, in Vanuatu). There, "Jacques" became "Saghi" (gh = voiced velar fricative), logically enough once you had noticed that a motor vehicle was "taraghi", from Pidgin "trak", itself from English "truck". I had a particularly sharp old man for my main informant. One day, I don't know what got over me, I asked me to "spell out" the Pidgin word "antap" ("on, above"). Without the slightest hesitation he said: a - ndra - vu "Antap", I believe, he had "stored" as "andravu", three perfectly good syllables of his mother tongue, which, in that language, would be pronounced as, indeed, "andravu", but which he *unfailingly* pronounced "antap" in Pidgin. (There is no such word as "andravu" in the Lolovuevue dialect or at least, I never came across one). I said an "important" piece of trivia because I think it reflects something we have wired-in, or close to it, and is very important and should constantly be kept in mind when we attempt to specify the desirable properties of a constructed language. I haven't given any further thought to the matter, to tell you the truth, and I do not know even vaguely what to make of it, nor how. Date: Tue, 1 Sep 92 01:00:46 -0400 From: lojbab@grebyn.com (Logical Language Group) Subject: Re: IAL desiderata, part 2 of 2 I've decided that I could ALMOST agree with Rick H's criteria for an IAL except for its apparent absolute-ness. He seems to regard certain things as being 'obviously' superior, based on criteria that he presumes are vital to an IAL. One of these for example, is that an IAL must have vocabulary that is 'easy to memorize'. This assumed criteria in turn dictates several other design features. Unfortunately, I, as I've too often said, disagree with this semi-hidden assumption. I think that a new language must be learned as a unit, as much as possible discarding semantics of your previous language as an assumption, and this is hard to do when you have cognates - so I regard cognates as a risk and not an ease. I am finding in studying Russian (as well as Lojban) that I understand things better when I am forced to think about them IN RUSSIAN, rather than be translating them back into English terms. (I also seem to do OK translating them back into Lojban terms, because this requires a reformulation and thus an understanding before I do the translation, but that is still in effect analyzing IN RUSSIAN at least partially.) I think Rick needs to bring his assumptions about what is good more out into the open before stating the lists of criteria that result from those assumptions. He should then for each criteria indicate which assumptions that criteria depend on. FInally, the article should make clear that others may have different assumptions which will of course result in different criteria. (He half does this at the beginning by stating that some points are debatable, but in wording that seems half-hearted.) Finally, and this I think is key for both conlangs in general as well as for IALs in particular among conlangs, he should make it clear that violating an assumtion (less likely) or a derived criteria (more likely/permissible) MAY be justified as a tradeoff to gain some other benefit. Thus for example, we consider Loglan/ Lojban's allomorphy to be justified sufficiently by factors including redundancy and variety in learning levels to more than make up for the (for us) minor benefit seen in 'ease of memorization'. (I should note byu the way that his example of allomorphy, the variation in English plural morpheme, is NOT The same as ours, in that for us, BOTH morphemes in an alternative are equally permissible given that they fit morphology rules and the resulting forms give the same meaning. The closest equivalent I know of is the purposrted alternation between VAXen and VAXes among hackers as a plural for VAX. This would not be a problem for people learning English, I claim, if -en could ALWAYS be alternated with -es following a consonant.) MY argument it would seem would be self-evident, in that, if there REALLY WERE any universal criteria for an IAL, then no doubt everyone who worked on IALs would come up with somthing virtually identical to the others. This doesn't happen, and the debates tend to be especially nasty (as with the accusative wars). To me this tells me that people who want to state principles of conlangs design should want to be very clear in stating that their criteria are either personal, or at least likely to be weighted by their own cultural/aesthetic values, and that people of other language backgrounds, for example, may justifiably have different criteria that are quite just as good. Since so many of these lists of criteria seem to end up with a predominance of English-language like criteria, and are predominantly written by English speakers (who are at least as biased towards their language as an aesthetic ideal as any), I would think broad-minded English-speaking conlang inventors like Rick would be bending over backwards to avoid being seen as Anglocentric. lojbab Date: Tue, 1 Sep 92 08:27:11 MDT From: mnu@inel.gov (Rick Morneau) Subject: re: IAL desiderata Rick Harrison writes: > > If "board" generally means "a processed piece of wood," > I consider the "board" in "board of directors" to be an undesirable > instance of polysemy > I don't think this is a good example of polysemy. Although the "board" in "board of directors" may have derived ultimately from the "wood" meaning, I think they have diverged enough to be considered homographs, rather than polysemes. Regardless, this would certainly fall into the category of stretching the meanings to the breaking point. I wrote: > > allomorphy may be OK for a loglan even though it is undesirable > for an IAL. > and Rick responded: > > Sure, it's okay for loglangs that don't claim to be suitable for > the role of international auxiliary language. > I suppose that this is where the real crux of the matter lies. However, since I would like to avoid yet another battle with the loglanists, I will say no more. Dona nobis pacem. :-) Finally, it pains me very much to agree with Jacques Guy, Richard Kennaway and Rick Harrison about the undesirability of vowel clusters (although diphthongs are OK by me). I've studied a little Hawaiian (and have since forgotten almost all of it), and I consider it to be one of the most beautiful languages I've ever heard. However, the lengthy vowel clusters ARE difficult to master, and I don't feel that they should be included in a conlang. Fortunately, I believe that one can achieve both beauty AND ease of learning. As it turns out, I know of another language that I feel is even more beautiful than Hawaiian, but which is extremely easy to pronounce - Swahili. It's syllable structure is essentially (C)(S)V(N), where S is a semivowel and N is a nasal, but I've never seen a word that contains more than two consecutive V's, thus avoiding the difficulty of languages like Rotokas and Hawaiian. Regards, Rick -- *=*= Disclaimer: The INEL does not speak for me and vice versa =*=* = Rick Morneau Idaho National Engineering Laboratory = * mnu@inel.gov Idaho Falls, Idaho 83415, USA * =*=*=*=*=*=*=*=*=*=*= NeXT Mail accepted here! *=*=*=*=*=*=*=*=*=*= From shoulson Tue Sep 1 09:47:40 1992 EDT Subject: In defense of grammatical gender Me, I think grammatical gender as masc/fem is more trouble than it's worth: makes for difficulties when speaking in general (even if you have a neutal pronoun), what do you use for animals (who definitely have a sex, but are usually called "it" anyway). Personally, I very much like the method used in Okrand's Klingon (and numerous natlangs as well, of course), wherein there is no distinction between male and female (which strikes me as a terribly trivial distinction to bake deeply into a language), but there *is* a distinction between sentient and non-sentient objects. Sentience is usually far more important to a speaker than the physiological make-up (or worse, arbitrarily assigned category) of the person/thing in question. The criteria suggested for sentience is ability to use language. A servicable definition, and broad enough to allow fudging for grey areas (there are always grey areas). That's actually one of the things I sort of miss in Lojban (not that it should be changed: it would break Lojban's model and world-view). Since it doesn't distinguish between sentient and non-sentient, there's no distinction between, say, "Who came here" and "what came here". Note that most natlangs which lack a neuter for use with objects still retain separate words for "who" and "what", the distinction usually being person (sentient) vs. thing (non-sentient). English-speakers would very likely be much less uncomfortable with collapsing "he" and "she" into one pronoun than they would collapsing "who" and "what". Even in modern parlance, people are averse to using "it" as a gender-neutral (as opposed to neuter) pronoun, as it somehow seems "insulting", though it wouldn't be too bad a stretch conceptually to go from neuter to neutral. If I were building my own conlang (something I probably ought to do, just for kicks) (with a different purpose than Lojban or Esperanto or anything, of course), I would put in pronouns for sentient and non-sentient as my "genders", along the lines of Okrand's "ghaH/'oH" and "chaH/bIH". ~mark From: cbmvax!snark.thyrsus.com!cowan@uunet.UU.NET (John Cowan) Subject: Conlang criteria: the very idea! Date: Tue, 1 Sep 92 13:03:19 EDT Disclaimer: The following discourse is meant abstractly. As is obvious, it was triggered by Rick H.'s list of criteria for conlangs, but nothing in it is meant as a reply to particulars nor as a denigration of that list. "They do but jest, poison in jest, no offence i' th' world." It seems to me that there is something inherently perverse and unreasonable about the idea of a list of criteria for conlangs. A conlang being itself a specification (for linguistic behavior), we are then presented with a specification for a specification. In what way does that differ from the conlang itself, except in lack of detail? Let me compare conlangs to another domain that will doubtless be familiar to conlangers: text editors. How should one go about writing a specification for an ideal text editor? Obviously it should be able to display and modify text: that is the analogue of the notion that a conlang should be able to be spoken, understood, read, and written. But by what means should these broad goals be achieved? The partisans of EMACS would undoubtedly list "redefinable commands" as an essential feature, and perhaps "modelessness". Vi users (at least English natives) would mention "mnemonic command families". Troglodytes like myself would demand a rich set of commands available from a command line (I am using 'ex' to type this very message). Can there be any doubt that the specification of the ideal editor would simply reflect the specification of one's own editor? "Never argue about religion, politics, or text editors." Now everyone has features in their favorite editor which could stand improvement. (I would like 'ex' to word wrap automatically, for example.) So the "specification for the specification" for my text editor would include a number of properties that I don't have now. Nevertheless, the underpinnings would be essentially those of 'ex'. What seems to me would be far more useful than a set of specifications (and I say this knowing that I don't have to write it, and am not equipped to) would be a set of engineering trade-offs. For example: "If your language does not mark case, then it will have problems with ellipsis unless the order is SVO." What other such generalizations can be made? -- John Cowan cowan@snark.thyrsus.com ...!uunet!cbmvax!snark!cowan e'osai ko sarji la lojban. Subject: Re: Conlang criteria: the very idea! From: Jim Gillogly Date: Tue, 01 Sep 92 16:59:14 PDT > cowan@snark.thyrsus.com (John Cowan) writes: > "They do but jest, poison in jest, no offence i' th' world." Boys throw stones at frogs in jest, but the frogs do not die in jest -- they die in earnest. [Heron, I think.] John raises some interesting points, and opens a fascinating area for exploration with his: > would be a set of engineering trade-offs. For example: "If your language > does not mark case, then it will have problems with ellipsis unless the > order is SVO." What other such generalizations can be made? and I'll eagerly await Jacques' musings on this subject. However, Rick H's proposal was not just an effort to provide "a list of criteria for conlangs", but specifically a list of desiderata for an IAL. That makes more sense than prescriptively telling people what they shouldn't have in their conlang design. Conlang designers have lots of goals, and designing an IAL may not even be the most common one. For me (a pessimist in the IAL battles) (excuse me, a realist), the question is not what features a conlang would need in order to be acceptable as an IAL, but rather what features are sure show-stoppers... and I think some of those can be identified. The next issue in my mind is whether anything is left as a potential IAL once all the show-stoppers have been identified. To go back to John's editor analogy, of *course* I won't tell you that you have to have auto-fill in your text editor. However, if it has only a line-at-a-time interface, I promise it won't be my International Auxiliary Editor unless I go blind and *need* line-at-a-time. Jim Gillogly Date: Tue, 1 Sep 1992 23:11:07 -0700 From: D Anton Sherwood Subject: ease of learning Rick Morneau says: > My personal feeling is that a credible IAL must be just as easy to > learn for a Chinese or an Indonesian as it is for an Arab or an > Italian. I disagree strongly. I prefer a minimax test. If Antonese is easier for the Indonesian than for the Italian, but easier for the Italian than Bahasa Indonesia itself, who is worse off? If Mornellic is harder for the Italian and the Indonesian than Esperanto is for either, who is better off? What's the virtue of equal difficulty? (For a Sapir-Whorf experimental language, I'd want maximal dissimilarity to the languages of the subjects, but that's not an IAL.) I doubt it's possible to design a language that's equally easy for everyone, anyway. Indeed it's hard to believe that _any_ language is equally easy for you and me. How do you normalize a standard member of a language community? Do you penalize those language communities with a high rate of bilingualism, because (I suppose without evidence) they have a natural advantage when it comes to learning a third language? Anton Sherwood = dasher@well.sf.ca.us From: "Edmund Grimley-Evans" Date: Wed, 2 Sep 92 13:20:03 +0200 To: conlang@buphy.bu.edu Subject: IAL desiderata: tones et al. > That said, it is a pity that there is much, very probably valid, > reluctance towards tones. Tones are good for speech recognition, I'm > told by a colleague here who's into it (and consonants are bad, > especially nasals and voiceless stops), and speech recognition is > fashionable. If your language were to consist *only* of tones, speech recognition would probably be absolutely trivial. You could speak to your computer tomorrow... Maybe a good conlang should have a standard encoding as a purely tone-language... Edmund From shoulson Wed Sep 2 09:58:58 1992 EDT Subject: IAL desiderata: tones et al. From: "Edmund Grimley-Evans" Date: Wed, 2 Sep 92 13:20:03 +0200 If your language were to consist *only* of tones, speech recognition would probably be absolutely trivial. You could speak to your computer tomorrow... Maybe a good conlang should have a standard encoding as a purely tone-language... Congratulations on discovering Solresol. *That* could be spoken purely as tones, if I recall. I wish I had some data on it; anyone out there have a brochure or anything? I have a few pointers from Rick H; has anyone got data as well (as opposed to pointers)? Correct me if I'm wrong, but Solresol's "alphabet" was really a syllabary, consisting of, yep, "do", "re", "mi", "fa", "sol", "la", and "si". You could sing it if you had perfect pitch or at least relative pitch and could tell your listener the tonic. A language entirely of tones. And I've seen hand-held devices for $30 or so that can tell you what note you're singing; how's *that* for demonstration of how much can be done with current technology? Somehow it doesn't strike me as easy to learn, though... ~mark Date: Wed, 2 Sep 92 07:31:13 MDT From: mnu@inel.gov (Rick Morneau) To: conlang@buphy.bu.edu Subject: Re: Conlang criteria: the very idea! John Cowan writes: > > It seems to me that there is something inherently perverse and > unreasonable about the idea of a list of criteria for conlangs. > A conlang being itself a specification (for linguistic behavior), > we are then presented with a specification for a specification. > In what way does that differ from the conlang itself, except in > lack of detail? > I would hardly call it a "specification". Listing features found in natural languages and discussing the pros and cons of whether they should be included in a conlang is NOT a specification. However, it has certainly been rich food for thought, and has shown us how much people's preferences can vary. (I NEVER would have expected a debate on tones! It suprised the hell out of me. It was great!) An additional benefit of Rick H's posts is that it has obviously got a lot of people thinking. The fact that people disagree is only to be expected, and can be (has been!) the springboard for some healthy and fun discussion. There's nothing perverse or unreasonable about Rick Harrison's posts. My opinion, of course. :-) > > What seems to me would be far more useful than a set of > specifications (and I say this knowing that I don't have to write > it, and am not equipped to) would be a set of engineering > trade-offs. For example: "If your language does not mark case, > then it will have problems with ellipsis unless the order is SVO." > What other such generalizations can be made? > Certainly, we need these trade-offs, but they can and should be used hand-in-hand with the items in Rick's articles. My own essays on conlang design discuss many of these trade-offs. Unfortunately, it's a simple fact of conlang life that people who have already adopted a particular conlang are NEVER going to like what Rick and I have to say. Whenever we say something that implies a flaw in an existing conlang, proponents of that conlang will become upset, as you have. :-( Regards, Rick -- *=*= Disclaimer: The INEL does not speak for me and vice versa =*=* = Rick Morneau Idaho National Engineering Laboratory = * mnu@inel.gov Idaho Falls, Idaho 83415, USA * =*=*=*=*=*=*=*=*=*=*= NeXT Mail accepted here! *=*=*=*=*=*=*=*=*=*= Date: Thu, 3 Sep 92 00:10:47 -0400 From: lojbab@grebyn.com (Logical Language Group) Subject: "easy to learn" Considering how many times that the question of what language is "easy to learn" for whom, is perhaps the most debated topic on sci.lang, I would say that it is far from obvious that any consensus could be reached on either what criteria determine "easy-to-learn-ness", and most important, what their relative priority is. These firefights almost always end with the conclusion that the question is not scientific, so that the answer cannot be either. I don't think you can talk about a language being 'easier' for an Indonesian or for an Italian, without making a lot of assumptions about what makes something easier. Though a monolingual English speaker, I am finding Russian much easier to learn as an adult than I found Spanish as a child, even though kids are supposed to learn more easily and I had a native Spaniard for a good friend. I've found the alphabet to be a trivial problem compared to even the most minor of semantics differences, but I have gotten the feeling that many Americans would say that Russian is a harder to learn language SOLELY on the basis of the alphabet. lojbab Date: Thu, 3 Sep 92 15:59:46 EST From: j.guy@trl.OZ.AU (Jacques Guy) Subject: Re: "easy to learn" The measure of learnability is in the learning, I should say. Not out of a textbook, because it depends so much on the quality of the book, not from a tutor who knows your language, because it depends so much on how good the tutor is, but parachuted into a monolingual community with no-one else speaking your language and you not a word of theirs. That's how I learnt Sakao, the topic of my thesis. Well, no matter whether I had been French, Italian, Indonesian, Korean, Hopi, a Klingon or a Hobbit, I would have had a much, much easier time of it learning Tolomako, which was also spoken in Port-Olry. And still easier learning Beach-la-mar, the Pidgin of the country. At the same time, I had an easier time learning Sakao than an Italian or an Indonesian would have had, because, with the 20 or so vowels of my variety of French, I had not very much trouble picking the 12 Sakao ones, even though they were slightly off the French ones. So there definitely are languages that are simpler than others, and by a long shot, too. If you don't believe me, just try learning Navaho, or French, for that matter. If learnability is one thing we are looking for in a conlang, we ought to examine those simple, that is, easily learnt, languages, and draw lessons from them. How do you tell them? Easy: round up all the Pidgins for which data is available. Beach-la-mar, New-Guinea Pidgin, Police Motu (an Austronesian-based Pidgin of New Guinea), Chinook Jargon, Sabir, go and tape the sound track of Dersu Uzala, where you can hear much Russian pidgin (I remember Dersu Uzala shaking his fist at the water vendor saying: "Tvaya, plahaya ludi!"). Let's round them up and ask ourselves: what have they got in common? And what is it that they don't have? So far, I haven't found one with tones, I haven't found one with cases, I can't think of one with inflected verbs (I'd have to check Police Motu, which I have completely forgotten). Haitian Creole has suffixes, like Arabic and Hebrew, to indicate possession, but they're transparent, being abbreviated forms of the pronouns, e.g. papam "my father" derived from papa-moi (would you believe that "mon papa a` moi" is perfectly good colloquial French?). But a creole is an adolescent pidgin, soon a full-fledged adult language. One thing everybody I think will at least agree with is: it shouldn't have clicks! From: And Rosta Subject: tones Date: Thu, 03 Sep 92 13:18:46 +0100 I feel there's a big difference between (a) Forcing every syllable to be either Hi or Lo tone and (b) Allowing whole words to be associated with a single tone, flat or contour, which spreads through the whole word however few or many syllables it has, or Allowing syllables to be associated with contour tones, or tones at four different pitch levels. (a) I could handle: it's not far off English stress. Japanese is like this. (b) is torture, but gives phonologists lots to be getting on with. --- And. Date: Thu, 3 Sep 92 07:32:30 MDT From: mnu@inel.gov (Rick Morneau) Subject: Re: "easy to learn" Howdy conlangers! This discussion about ease-of-learning bothers me because people are making definitive (and wrong!) statements with nothing to back them up. Instead, they're making up truisms that may seem logical at first glance, but which simply end up reflecting their personal preferences. As it turns out, ease-of-learning is definitely measurable. A LOT of work has been done that, although not aimed at conlang design, can certainly be applied to judging whether a language is easy to learn. As one of many examples of such work, several studies have been sponsored by the Foreign Service Institute of the U.S. State Department about the problems encountered by teachers in teaching certain languages to students of different linguistic backgrounds. For instance, Chinese students have a difficult time with English tenses, because they are not used to having to indicate tense all the time (in Chinese, tense is usually indicated only when it cannot be determined from context). Conversely, English speakers have difficulties with Chinese tones, because they are not used to using pitch contours to make phonemic distinctions. Yet English speakers have no difficulty at all in learning to use verbs without marking them for tense. Language educators would agree unanimously and without reservation that it always comes down to this: a language becomes more difficult to learn WHEN THE STUDENT MUST LEARN TO MAKE DISTINCTIONS THAT HE IS NOT USED TO MAKING. That's why students have difficulties with things like accusatives, mandatory tenses, tones, noun/adjective agreement, honorific inflection, consonant and vowel harmony, mandatory gender distinctions, polysynthesism, ad nauseam. And that's also why a language like Indonesian is so easy to learn for EVERYONE because it does not have any of these obstacles. Jacques Guy writes: > > So there definitely are languages that are simpler than others, and > by a long shot, too. If you don't believe me, just try learning > Navaho, or French, for that matter. > Hallelujah! Regards, Rick -- *=*= Disclaimer: The INEL does not speak for me and vice versa =*=* = Rick Morneau Idaho National Engineering Laboratory = * mnu@inel.gov Idaho Falls, Idaho 83415, USA * =*=*=*=*=*=*=*=*=*=*= NeXT Mail accepted here! *=*=*=*=*=*=*=*=*=*= Date: Mon, 14 Sep 92 09:45:26 EST From: j.guy@trl.OZ.AU (Jacques Guy) Subject: phoneme selection; Neanderthalese There can't be anything universal there I'm afraid short of a dreadfully useless inventory: three vowels at most and four or five consonants. Anton Sherwood asks: Are there any languages with only one nasal? *&#%#! Rotokas hasn't got any! Viz: g = voiced velar fricative v = voiced bilabial fricative r = alveolar flap t = voiceless dental stop, affricate when followed by i k = voiceless velar stop p = voiceless bilabial stop and that's it. Fortunately, Rotokas speakers don't pull much weight on the international scene. Incidentally, this reminds me of the theory according to which Neanderthals had no language because their vocal tracts could not produce a great enough variety of sounds. Since the Rotokas have been getting along with only 35 different *syllables* (counting CV or V as a syllable), the argument seems very specious to me. Date: Tue, 15 Sep 92 12:18:48 EST From: j.guy@trl.OZ.AU (Jacques Guy) Subject: SVO and embedding in a very roundabout way dfihueg@rz.uni-sb.de's question about natural languages with a near Polish or reverse Polish grammar, Rick Morneau's draft on AL design, some of my own cogitations, plus memories of Bichelamar, led me into some thoughts. It's a meandering, attempt at dreaming up a conlang just to see. In the process I came up with a possible reason why SVO is so frequent, and something else (about embedding) which I am still not sure what to make of. Here goes... Consider this rather degenerate tree: Agent-----Action------Victim E.g. He----- broke ----the window Let's pick up this tree by the agent. Dangling from the agent, we have the two nodes: action and victim, in that order. In other words, an SVO sentence. If we pick it up by the victim we have an OVS sentence. Now, all verbs [note 1] have an agent, but not all have a victim (i.e. not all are transitive). In order words, we can always pick up a sentence by the agent, but not always by the victim. Could that be one of the reasons for the predominance of SVO and the rarity of OVS? Next, let's try to pick this sentence up by the action node. We have two nodes dangling and that won't do. It would do if we had two mouths and two sets of lungs with which to articulate both at once but, as our anatomies go, we've got to make a choice, and that is where the problems start. Being fundamentally lazy and shy of problems, I threw this verb-first approach into the too-hard basket in the fervent hope that it would just go away. Next, I looked at a leafier tree: Agent-----Action------Victim | -------Instrument E.g. He----- broke ----the window | -----with his ball Alas, the problem has not gone away! No matter which end I grab, I end up with two or three dangling leaves, each shouting "Pick me, pick ME, PICK ME!!!". Then I look at this little word "his" and shivers run down my spine. After some head-scratching, I cowardly decide to attack the little fellow first. "His". "His ball". What's that? "He"+"ball" Ha, ha! I'm going to invent a possessive binary operator! "Broke with his ball" == "Broke" ^ "his ball". Instrumental binary operator! ... what a mess! Suck back, Frogguy. The little fellow was no wimp, tackle him another way. "His ball" == "the ball that he owns". That is nice, because it looks as if I've reduced the problem to finding a way of expressing relative clauses, and if I can do that, it will be a big chunk of the language, bigger than a mere possessive case. Bichelamar to the rescue! But let's just use English glosses. I'll only show that a verb is transitive by suffixing "'im" to it. "The ball that he owns" --> "ball which he own'im". Have I got the little fellow licked yet? Better check before I move on to the big fellow "Instrumental". "The police caught the man who robbed the bank" --> Police catch'im man which rob'im bank Suddenly I realize that I have forgotten something in my Bichelamar- like English: Ball which he own'im HIM ^ | | | ------------------- Man which HIM rob'im bank ^ | | | -------- Thinking back on those languages which two, three, four and more possessive relationships, I realize that the Bichenglish way of expressing possession is even richer. Perhaps he didn't *own* the ball, or perhaps again he did. We can express that saying that he had the ball, or Russian and Celtic-style, that the ball was with him: Ball which he have'im HIM Ball which HIM with'im he Let's see... 1. Police catch'im man which him rob'im bank 2. Ball which he have'im him 3. Ball which him with'im he "Him" refers to the noun immediately preceding "which". "He" refers to another noun, but which? It occurs only in (1) and (2) which are not complete sentences. Must complete them. That was "He broke the window with his ball". We have two bits: He break'im window Ball which he have'im him Put them together, avoiding inventing new parts of speech: He use'im ball which he have'im him, he break'im window. "Him" refers to "ball", "he" to the subject of "use", which is also the subject of "break". This is the place where to tread very, very carefully: I sense a trap somewhere. He use'im ball, he break'im window. Fine. You use'im ball, ... he break'im window? you break'im window? Red alert! In Englamar (sounds better after all than "Bichenglish"), "him" refers to the noun immediately preceding "which", and, so far, "he" has referred to the subject of the preceding verb. "Use'im" and "break'im" share a common subject. This call for coordination, and I'll happily and remorselessly steal Rick Morneau's idea, using three different words or expressions for each of our "and" and "or", normalizing the perfectly good, but verbose, English "not only... but also... and... and... and finally...", i.e. in Englamar: and: "both...", "and ...", "and-finally..." or: "either...", "or ...", "or-finally..." Thus: "He both use'im ball and-finally break'im window." And: "He both use'im ball which he have'im him and-finally break'im window." At this stage we have a language that allows infinite embedding. Well, my poor brain can't cope with double embedding, never speak of infinite. I don't see much point in a grammar that allows structures that cannot be processed by human beings. Those are bugs, not features. I'm afraid that will have to go. Let it be plain: "He use'im ball and break'im window" What does that mean for Englamar grammar? A conjunction connects the word immediately following to the previous word of the same category: verb to verb, noun to noun. If that is the case, we cannot translate "he broke the window with his ball" as "he use'im ball which he have'im him and break'im window", for "break'im" would then connect with "have'im", not "use'im". Well, this one's rather easy for a change. Since "he" refers to the subject of the previous verb, we say: "He use'im ball which he have'im him he break'im window." Hm. Let's try this for size: The ball which his father bought him and with which he broke the window. I'm sure we are going to end up in a pretty mess. First, I must get my bearings. How would I say that in Bichelamar? Bol ya, we papa blong hem i kivim long hem... bang! stuck. "...mo we hem i brekem windo long hem" is whiteman's Bichelamar. What you ought to say is something that translates back into English as "That ball which his father bought him, he broke the window with it. That ball...". Never mind, let's see what Englamar can do with the rules painfully dug out so far: "Ball which father which him begat'im he(?)..." stuck for the noo! If I say "... and which" that last "which" connects to the previous one, the one after "father", but I want it to connect to the "which" after "ball". But come again: that is how it should be, for this sort of deep embedding I am after is not naturally processable by human beings, no more than "the wheat that the rat that the cat chased ate...". What I was trying to say was: ---------- | | v | Ball which father which him begat'im he x which he use'im him he break'im... ^ | | | -------------------------------------------------------- Same mess as: --------------------- | | v | The wheat that the rat that the cat chased ate cost tuppence a bushel ^ | | | --------------------------------------- But note how Bichelamar got us farther than Englamar. It is not that Bichelamar allows deeper embedding, it's that it has a possessive construction where Englamar resorts to a relative clause. Could it be that we have several "stacks", one for relative clauses, one for possessives, one for..., each one quite shallow, allowing only single embedding? I tend to think so. What if Bichelamar didn't have a possessive construction? How could I say that sentence? Easy: "Papa we i bonem hem i kivim wan bol long hem. Be hem i brekem windo wetem bol ya..." (bonem = to beget, give birth to, lay [eggs] <-- "born" + the transitive suffix -em). Literally: "The father who begat him gave him a ball. But he, he broke the window with that ball" [note 2] Or again: "He broke the window with a ball. The father who begat him had given him that ball" (Hem i brekem windo wetem wan bol. Papa we i bonem hem i kivim bol ya pastaem). Except that in that last case, in Bichelamar at least (I'm not fluent in Englamar, never will be), it sounds as if the kid's father had given him a ball so that he could break the window with it. In the first version (Papa... Be hem...) it is quite clear that that was not the intent. At this stage it is appropriate to pause and reflect on the wisdom of the inhabitants of those not-so-very-long-ago cannibal islands. For "the father who begat him" is silly. "Father" implies "beget" and children are what make a father, so there is no such thing as "father" only "father of (someone)". Ditto for body parts. When is it you last saw a stomach taking a stroll on its own? They carry this logic over to verbs. There is one verb "to eat" and one "to eat (something)". But how, will you ask, would you say "a father"? Can you say it at all? Yes: "someone's father" if you insist. Still, it sounds... as if you were not quite the full quid, talking about someone in terms of his children whom you don't seem to know, nor care to. What? Those wretched savages have no language in which to say "Dads and mums"? Oh yes, but not like that. Like this: "men and women, when children are born to them..." I keep telling you: those languages were bullshit-proof. Now I become dimly aware that there is a class of nouns which are to other nouns like transitive verbs are to other verbs. Kinship terms are such, body parts too. I'll call them... transitive nouns! (Those shrieks of indignation are water off a frog's feathers for I'm getting deaf in my middle age). What's the big fuss about, anyway? If I'm a father, it's because I've fathered a child, isn't it? Abstract the tense, and now tell me what difference there is between "John is Joan's father" and "John has fathered Joan". Zilch. Make it Englemar "John papa'im Joan". Perhaps I should leave Bichelamar to recover from this hefty idea transfusion and turn to another pet language of mine: Ancient Chinese. Take an intransitive verb, any intransitive verb, or an adjective, any adjective, and use it as if it were a transitive verb. Like this: take "red". "I red" means "I am red". "I red town" means "I cause the town to be red" or "I consider/think/see the town as red". Sure, it's ambiguous, but it's elegant and economical. And economy of words is something Bichelamar and Englamar could well do with, don't you think? Can we transport that into Englamar? Let's see. If X A Y means X cause Y to be A, this implies: 1) X A meaning X is A 2) A Y meaning Y is A For instance: 1) John dead = John is dead John dead Jim = John kills Jim 2) Dead Jim = Jim is dead John dead Jim = John kills Jim Time to stop and think. I have a gut feeling that I have missed something very interesting and useful. Must backtrack. Some other time, though. Just one more thought before signing off. You remember that "eat (intransitive)" vs "eat (transitive)" of the savage islands? English has the same, except that the two verbs are unsullied by any overt differences: feed (intr) and feed (trans), viz: I feed the cat. The cat is feeding. That is very much like ancient Chinese. "Feed (intr)" is the basic form. If I add an object, "feed" becomes causative, i.e.: "I feed the cat" = I cause the cat to feed. I won't go into "I feed the mouse to the cat"... yet. Footnotes [note 1] Not strictly all. One may have to consider a very few verbs such as agentless, e.g. "to rain", "to snow" etc. Yes, you could say that the subject is "it", i.e. "the weather". Could be true. Could also be a kludge. [note 2] "Be" is a cross between English "but" and French "mais". If you want to sound educated, you'll use "bat" instead. The old pagans from up in the mountains will mark you for one of those &^%$! ^*@%#s. Why did I use "but"? It felt right. With "be" I introduced an adversative meaning. Here the adversative meaning is between the subjects of the two sentences: Papa..... Be HEM.... "Hem" is not "papa". Like in English "He did this, but he did that". The two he's are not the same person. From: "Edmund Grimley-Evans" Date: Tue, 22 Sep 92 15:29:31 +0200 Subject: unambiguous grammar for a conlang There wasn't much of a reaction to my proposal about reverse Polish grammars for making PP-constructions unambiguous. I had expected more of a reaction, because some conlang-designers think that a good conlang should be syntactically unambiguous. What some of these conlang-designers then do is to add "unnatural" end-of-phrase markers to a traditional grammar. I suspect if anyone started actually speaking such a conlang then they would immediately stop using those end-of-phrase markers, whether of not the official grammar allows them to be left out. Of course I'm not claiming that reverse Polish is more "natural" than end-of-phrase markers. What I want to know is: Is there a "natural" way for a human language to be syntactically unambiguous? Edmund From: maxwell@ltb.bso.nl Subject: "natural" syntactic unambiguity Date: Tue, 22 Sep 92 17:07:08 MET DST Edmund poses the question as to whether there is a natural way for languages to be syntactically unambiguous. Well, there seem to be natural ways to avoid *some* types of syntactic ambiguity. The classic example is casemarking in a language with free word order. Without casemarking there would be syntactic ambiguity about the functions of many NPs. But there are some kinds of ambiguity, such as that involving attachment of PPs, for which it is not so easy to find a natural solution. Lojban and probably its loglang relatives have a solution, but is it natural? Rick M. has already criticized Lojban for using constructions not found in any natural languages. One might argue that these solutions are unnatural on the same basis, although they probably seem considerably less natural than reverse Polish, at least to someone who doesn't derive his/her views from what is found in nature. Lojban's solution to several types of syntactic ambiguity involves counting arguments: the first is the agent, the second is the patient, etc. One can argue that counting is natural: I don't know of any language that doesn't have any numbers at all, although some of them don't go beyond three, I have heard, and a primary function of numbers is to count. So if counting is natural, maybe it is not so unnatural to use counting as a way of resolving ambiguity. This is certainly true, for computers, the counters par excellence. For human beings the evidence has to be more indirect, if it exists at all. I remember a paper many years ago by Alan Bell of the U. of Colorado with the title, "If speakers can't count syllables, what can they do?" I never read the paper, but the author evidently argues that counting syllables is an important part of some phonological rules. This is hard to doubt, since plenty of languages have a stress rule involving the Nth syllable of the word: first, last, next-to-last,... But I don't know of any evidence like this that counting is used to determine where to attach a PP or otherwise determine argument structure within a clause. Dan Maxwell@ltb.bso.nl -- From: huijsen@cs.utexas.edu (Willem-Olaf Huijsen) Date: Tue, 22 Sep 1992 16:01:29 -0500 Subject: Counting In response to Dan Maxwell: In your response to Edmund, you mention all languages having numbers, and some not going further than three: > Lojban's solution to several types of syntactic ambiguity involves counting > arguments: the first is the agent, the second is the patient, etc. One can > argue that counting is natural: I don't know of any language that doesn't have > any numbers at all, although some of them don't go beyond three, I have heard, > and a primary function of numbers is to count. So if counting is natural, > maybe it is not so unnatural to use counting as a way of resolving ambiguity. > This is certainly true, for computers, the counters par excellence. For human > beings the evidence has to be more indirect, if it exists at all. In the television program 'Before Bable' of the BBC (England), a rather superficial excursion is made into the research on the alledged common origin of *all* languages. One of the subjects discussed is numbers. Comparing numbers in several languages, a common form for 'one' and 'two' can be found, but for 'three' this is not the case. The program suggests that this may be because the oldest language(s) did not have numbers beyond two. A small remark: as you may know, an, in my opinion beautiful, word for the next-to-last syllable is the penultimate syllable. Greetings, Wolf Willem-Olaf Huijsen huijsen@cs.utwente.nl Date: Wed, 23 Sep 92 08:21:11 EST From: j.guy@trl.OZ.AU (Jacques Guy) Subject: Numbers!? What? Who are those silly TV people who've found a common form for 'one' and 'two'? Or have they counted Japanese as fallen out of outer space? One= hitotsu (<-- *pito), two = futatsu (<-- *puto). And Swahili? -moja = one, -wili = two. And... and... *choke* *gasp* *gag* Date: Wed, 23 Sep 92 09:40:30 EST From: j.guy@trl.OZ.AU (Jacques Guy) Subject: syllable counting in Nasioi Dan Maxwell metioned a paper by one Alan Bell (who's that?) by the title "If speakers can't count syllables, what can they do?". Fasten your seat-belts again. In the Nasioi language (Papa-New Guinea) verbs takes prefixes to express person, number, I forgot what else. What I did not forget, though, is that the first CV syllable of a verbal prefix is reduplicated, apparently randomly. "Apparently" only, because there is a rule: it is reduplicated (or it isn't) so that the verb ends up having an even (or was it odd?) number of syllables. I know, it's disappointing that I can't remember much, it's so long ago. It was in 1969, during a crash course in fieldwork methods, Pike's phonemics, and tagmemic grammar. The Nasioi data was part of one of the many practical exercises we went through. No-one cracked the mystery of the Nasioi reduplicated syllable. Our tutor, who had studied Nasioi in the field commented: "I'm not surprised. It took us months to figure that one out." Not a feature I'd recommend for a conlang, though. How do the Nasioi manage that feat? Date: Fri, 25 Sep 92 07:35:22 EST From: j.guy@trl.OZ.AU (Jacques Guy) Subject: Languages where when=if=who=which=that This is another "plug", if you please, for the simplicity, elegance, and logic of those languages of Vanuatu which I have been so fond of quoting as paragons of what a universal language should be. The word for "if" is "va" in Tolomako. So is the word for "when" (the conjunction, not the interrogative adverb). Finally, to add insult to injury, Tolomako "va" translates our relative pronouns "who", "whom", "whose", "which", "that". And I call that simple, elegant, and logical? Tolomako verbs have no tenses, no aspects, only two moods, realis and irrealis. The realis corresponds fairly well to our past and present, the irrealis to our future, subjunctive, optative and imperative. In other words the realis is for what is granted, acquired, achieved, the irrealis covers what could be or could have been. "Va", with a verb in the irrealis expresses a contrary-to- fact condition, e.g. "if pigs had wings", "if the sea were boiling hot". With a verb in the realis it means "if" when the context indicates the future, and "when" when it indicates the past. There is no room for the present. And why should there be for "if" and "when"? (Think about it). Logical? Oh yes indeed. Say the context indicates the future. How dare you say "when"? "When" implies certainty, "if" possibility. Are you an infallible prophet that you should be allowed to say "when" referring to the future? Say the context indicates the past. What do you mean "if"? Either it happened or it didn't. If it did not happen and you are fantasizing about what could have taken place if it had happened, then "va" and the irrealis (contrary-to-fact condition) is what you ought to use. If it did happen, then there is no question of doubt, no conditional meaning [Note 1]. So it must be "when". How about "va" doing the job of a relative pronoun now? How can it ever be justified on logical grounds? First, let's clear the ground about relative clauses. Take this famous example out of la grammaire de Port-Royal: "the invisible God has created the visible world" ("Dieu invisible a cre'e' le monde visible"). It is glossed therein as "God, who is invisible, has created the world, which is visible" then as "God is invisible. He created the world. The world is visible". And that last gloss is precisely what you would say in Tolomako, or Sakao, or any of those pet languages of mine. But the English wording allows us to interpret the original sentence as "the invisible God (there's another one, visible) has created the visible world (there's another one, invisible)" i.e. "the God who is visible has created the world which is visible". In English the difference is only one of punctuation. In Tolomako it is one of syntax. You would say "this God "va" you cannot see has created that world "va" you can see". The verbs (can see, cannot see) would be in the realis. Logical? Sensible? I say YES! "This God" contrasts with another one (that God). Ditto for "world". Why? How do you tell this God from that God? *IF* you can see it it's this one, *IF* you can't it's the other(s). Sakao behaves in a similar manner. Only immensely more complex as it has seven degrees of deixis where Tolomako has only three (here/this, yonder/yon, there/that), and has many classes of deictics where Tolomako has only one. The two moods (realis and irrealis) which you have seen affect verbs also affect substantives. That leads to very interesting distinctions that English just cannot express cleanly. [Note 1] Not quite true. Say you are a chieftain involved in a tribal war right now. You have no news of Tari, one of your lieutenants, sent to attack Thngaru's rear. How do you express "if Tari won"? Alas, I was too involved in trying to understand Sakao phonology and holophrastic constructions to even think of that question to put to my informant, but I know in my inner brain that it would be "va" and the realis. Date: Fri, 25 Sep 92 11:23:21 EST From: j.guy@trl.OZ.AU (Jacques Guy) Subject: One, two... (was: Numbers) I still can't get over that story of human languages sharing the same words for "one" and "two", and since I have just developed a king-size headache and learnt not to fiddle with 600 megabytes of my employer and provider's data when not fully functional, I am killing time waiting for our weekly buzilantsh ritual. Serendipitously, I found, without even looking for it, Georges Ifrah's "From One to Zero -- A Universal History of Numbers" (Viking Penguin). He gives a few instances of those languages that stop at two. Aranda (Australian): one = ninta, two = tara Torres Strait (between New Guinea and Australia): one = netat, two = neis. So much for the legend that "one" and "two" are expressed by the same words the world over. There's another legend that gets killed off in there: that those one-and-two-only people cannot count very far. On page 11 you have a drawing showing how those very Torres Straits Islanders count up to 33. By matching body parts in orderly fashion. You start with the little finger of your right hand, move to the ring finger, then the middle finger, the index finger, the thumb, the wrist, the elbow, the shoulder, the sternum, continue left: left shoulder, left elbow, etc... ending with the pinkie. That takes you to 17. Then the little toe of your left foot, next toe, ... big toe, ankle, knee, left hip: 25 so far. Switch to your right hip and continue until you reach the little toe of your right foot: 33! So you can imagine this conversation: "How many pigs do you own, Epeli?" "Left knee. No, sorry, I bought one the other day from Timoti; left hip". Overleaf are two other body-part counting schemes, both from Papua-New Guinea. My late colleague Donald Laycock once told me that the most he had encountered was a 47-point count. He also mentioned to me that many Australian Aboriginal languages only seemed to have words only up to two. The case, however, was that they had several full sets of numbers, each used for counting different things: you'd count, say, emus with one set and kangaroos with another. The "universal one and two" theory, however, presents a nice puzzle for puzzle-lovers. Namely, how did those people manage to select the right set of data to prop up their claim? I've tried, and failed. I had to jettison Chinese (Ancient: yat, nzi), Japanese (pito, puto), Swahili (moja, wili), Aztec (ce, ome), Maya (hun, ca) and many more that showed nothing like "unus, duo". I kept languages like Indonesian on the strength of "dua" (two), and, by necessity, its relatives, the other Malayo-Polynesian languages. But, but... those languages all disagree about the word for "one"! In Indonesian it's "suatu", in Tolomako it's "tea", in Tahitian "tahi" (unrelated to "tea"), in Fijian it's ... "dua"! Not only that, but most *agreed* with Indo-European for "three" and "four". Just consider: most have "telu" or a reflex thereof for "three" and "pat" for "four". And if we consider their "dua" related to, say, Latin "duo", it's hard to claim that "telu" is unrelated to "ter" or "tres", and "pat" to "quater" or "quattuor". Oh, here's Sumerian, fresh out of Georges Ifrah's book: 1 gesh, ash, or dish 2 min 3 esh 4 limmu 5 ia' 6 a`sh 7 imin 8 ussu 9 ilimmu 10 u Well, another one that just won't fit the theory very neatly. However hard I try, every time I find a language that matches Indo-European "one" or "two", either it also matches some numbers beyond that, or it's just "one" but not "two", or vice-versa. How, how did those people ever manage to doctor the data successfully? From: And Rosta Subject: Jacques's queries re proto-World numbers Date: Fri, 25 Sep 92 16:42:39 +0100 The BBC TV program mentioned earlier on Proto World was at the beginning of this year savaged on Linguist List. Although at that time none of the savagers (not _savages_) had seen the prog, it turned out they were justified. The prog consisted entirely of the speculations of a number of Proto-Worldists, & featured Greenberg, Ruhlen & a number of ex-Soviet Nostraticists (I forget the names - Dolgopolsky?). The claim reported in the BBC prog is that PW had numbers 1, 2 & 3. One - meaning also 'finger' is _tik_. I forget what the other two are. Thesort of counterevidence mustered by Jacques would be not faze the advocates of the PW hypothesis. Their claim is that if you consider all the lgs of the world you will find numerous geographically & genetically disparate lgs that all have very similar words for say 'milk/breast'. Searches of this sort yield a number of vocab items that are widespread & believed to have survived uunchanged from PW. I reckon the safest atttitude to these claims is that they are speculations for which the evidence is not very impressive but which are equually hard to falsify. A recentedition of Ruhlen's Aguide tothe World's Languages. vol 1 Classification_ discusses the above from ofcourse ahighly partisan perspective. --- And From: huijsen@cs.utexas.edu (Willem-Olaf Huijsen) Date: Sun, 27 Sep 1992 18:57:25 -0500 Subject: On Glaugnea Hello conlangers (shoudn't it be: collangers?), In number 16 of his Journal of Planned Languages, Rick Harrison publishes an article "A Glimps of Glaugnea" about a language "Glaugnea". A langue close, i.e. a language invented solely for the pleasure of the inventor. There Rick writes: ... Michael believes that standardized dictionaries and mass media hegemony tend to create an atmosphere in which linguistic innovation is discouraged; but there's plenty of reason to encourage individuals to engage in word creation and language invention. "Aside from AIRWOH (the allure of the extremely rare), and BLIXTH (something done for the beauty of the act, not for any inherent meaningfulness or utility) for the WOIGN [woin] (sole speaker of a language), there is one substantial reason for glossopoeia (word creation - red.) to be cultivated: not everything has yet been NAMED." Only the roughest outlines of our experiences are communicated by existing languages, so there is room for countless "interstitial languages" whose semantic structures need not coincide with those of any existing language. ... I support the idea that "not everything has been NAMED yet". Especially when creating both a language and a culture, the need for glossopoeia is obvious. Michael Helsem, the creator, has gone to quite an extent in glossopoeia. Here are some examples: welsha = the love of language snefk = to recognize a face and forget the name chabd = very effective for some, but not at all for others meebzink = the squeak of clean, damp hair lithonia = the beauty of something broken and now useless The point I want to make is that one should be careful in creating these words: there is no use in creating words for concepts that do not belong to the world(s) of experience of the user(s) (other than the AIRWOH and BLIXTH), since only words that are actually *used* deserve to be part of a language. For example, the last two words (above) are doubtful, in my mind. When I apply glossopoeia in my langue close (Umegian), I try to only create words when the need arises, i.e. when I frequently encounter a situation in which I need too many words for one concept too often. Greetings, The Wolf Willem-Olaf Huijsen huijsen@cs.utexas.edu From: maxwell@ltb.bso.nl Subject: Naming Date: Mon, 28 Sep 92 10:57:12 MET Willem-Olaf raises the question of what things should be named and suggests that the world of experience provides the answer. Things which are not fairly clearly part of this world are not important enough to deserve a name. For the purpose of new IALs (which are only a subset of conlangs), I would probably choose to be even more conservative: things which do not receive a name fairly often in natlangs are not important enough to receive a name of their own in a new IAL. If the culture of the IAL develops, then other more specialized and less frequent names can be added. But since IALs are supposed to be easy to learn, you don't want to burden the learners with esoteric names at least until the language has established itself. But for a conlang like Glaughea, which is not a IAL but has been created only for the pleasure of the creator, and perhaps others who share the creator's tastes, I see no reason to impose this restriction. Such languages, like science fiction, can also open our minds to possibilities that are untouched in natlangs and IALs. And who knows, maybe some of these possibilities will become reality one day. Dan Maxwell@ltb.bso.nl