nl-kr-request@CS.ROCHESTER.EDU (NL-KR Moderator Brad Miller) (06/04/88)
NL-KR Digest (6/03/88 17:07:39) Volume 4 Number 55 Today's Topics: Learning From Positive Examples DLT Book Announcement Bibliography Parser What are grammars (for)? Re: 3rd person pronouns Benjamin Lee Whorf Re: Observation ("Back to the Future"?) Submissions: NL-KR@CS.ROCHESTER.EDU Requests, policy: NL-KR-REQUEST@CS.ROCHESTER.EDU ---------------------------------------------------------------------- Date: Wed, 25 May 88 01:46 EDT From: Ken Laws <LAWS@IU.AI.SRI.COM> Subject: Learning From Positive Examples There seems to be agreement among some of the "What is a Grammar For?" participants that children can't learn grammars from postive examples alone. I hold that the proposition is false. One can learn a grammar (or at least competence, which I take to be equivalent) if 1) all legal constructions are in the training set (which everyone rightly rejects in this case), or 2) sufficient structure/similarity can be discerned among the language components so that all legal >>types<< of constructions are in the training set. In the latter case, one infers that constructions never encountered are not permitted. As an example, consider learning to cook from studying recipes. This may be more difficult than learning from a teacher, but it is not impossible. Each recipe is a local optimum, surrounded by a space of less desirable recipes (e.g., those with an extra tablespoon of salt). No one ever prints the negative examples, but with reasonable study one can infer their existence and many of their properties. -- Ken Laws ------------------------------ Date: Wed, 1 Jun 88 03:53 EDT From: Klaus Schubert <mcvax!dlt1!schubert@uunet.UU.NET> Subject: DLT Book Announcement MENT * BOOK ANNOUNCEMENT * BOOK ANNOUNCEMENT * BOOK ANNOUNCEMENT * BOOK ANNOUNC A new book in the series "Distributed Language Translation" went to the publisher in May and is due to appear not later than mid-August, before the Budapest conferences "New Directions in Machine Translation" and "Coling '88". B. C. Papegaaij / Klaus Schubert TEXT COHERENCE IN TRANSLATION (Distributed Language Translation 3) Dordrecht / Providence: Foris Publishers 211 pp. The book can be ordered from bookshops and directly from the publisher: Foris Publications, Postbus 509, NL-3300 AM Dordrecht, Netherlands; Foris Publications USA, P.O.Box 5904, Providence, RI 02903, USA; agent for Japan: Sanseido Bookstore, 1-1 Jimbocho, Kanda Chiyoda-ku, Tokyo 101, Japan CONTENTS Foreword 1. Text coherence in machine translation An appraisal of the problem and some prerequisites for its solution 1.1. Text and context 1.2. Text linguistics at a glance 1.3. Distributed Language Translation 1.4. Some terms and concepts of grammar 2. Clues and devices of text coherence A practical analysis 2.1. English to Esperanto: a sample translation 2.1.1. The text 2.1.2. Circumstantial information 2.1.3. A superstructure of text 2.1.4. A close-up of some translation steps 2.1.5. Conceptual proximity 2.1.6. Tense, mood, aspect and voice 2.1.7. Adding explicitness 2.1.8. Theme and rheme, repetition of words and syntagmata 2.1.9. Pronominal reference 2.1.10. Names, terms, abbreviations 2.1.11. Voice and markedness 2.1.12. Prominence and deep-case structure 2.1.13. Rhetorical patterns 2.1.14. Stylistic changes 2.1.15. Conclusion 2.2. Deictic reference 2.2.1. Pronouns and translation 2.2.2. Pronouns and text coherence 2.2.3. Syntactic restrictions 2.2.4. Syntactic redundancy 2.2.5. Syntactic rules and semantic features 2.2.6. The inadequacy of syntactic and semantic features 2.2.7. Focus 2.2.8. Speech acts 2.2.9. Conclusion 2.3. Content word reference 2.3.1. Lexical variation 2.3.2. Reference identity and translation 2.3.3. Indefinite versus definite reference 2.3.4. Directing the search 2.3.5. Definitesness without modification 2.3.6. Identity of sense or reference 2.3.7. Types of definite reference 2.3.8. Searching for reference 2.3.9. Conclusion 2.4. Communicative functions: theme and rheme 2.4.1. Thematic progression 2.4.2. Analyzing thematic progression 2.4.3. Given and new 2.4.4. Focusing rules 2.4.5. Breaking expectations, reintroduction of concepts 2.4.6. The complete pattern 2.4.7. Thematic patterns as a summary mechanism 2.4.8. The complete pattern, continued 2.4.9. Conclusion 2.5. Reduction to verbal elements 2.5.1. Case grammar 2.5.2. The verb as central element 2.5.3. Nominalized verbal elements 2.5.4. Some sample analyses 2.5.5. Using extended expectations to solve definite reference 2.5.6. Coherence through "deep" reference 2.5.7. Translating metaphorical expressions 2.5.8. When is a word a verb? 2.5.9. Finding parallel structures 2.5.10. Conclusion 2.6. Rhetorical structures and logical connections 2.6.1. Rhetorical patterns 2.6.2. Logical connectives 2.6.3. Conclusion 2.7. Towards the translation of text coherence 3. Text coherence within translation grammar 3.1. On the translatability of language units 3.2. Coherence of entities 3.2.1. Lexeme choice 3.2.2. Deixis and reference 3.3. Coherence of focus 3.3.1. How "free" is word order? 3.3.2. Communicative syntagma ordering 3.3.3. Grammatical features determined by syntagma ordering 3.3.4. Excursus on word order in dependency syntax 3.4. Coherence of events 3.5. Pragmatics - an escape from text grammar? 3.5.1. The kind of translation-relevant pragmatic knowledge 3.5.2. Full understanding? 3.5.3. Knowledge representation 3.6. Towards implementation Index References MENT * BOOK ANNOUNCEMENT * BOOK ANNOUNCEMENT * BOOK ANNOUNCEMENT * BOOK ANNOUNC ------------------------------ Date: Wed, 25 May 88 01:57 EDT From: Ken Laws <LAWS@IU.AI.SRI.COM> Subject: Bibliography Parser I have mentioned the possibility of a bibliography parser to AIList (and possibly IRList) as one of the great unsolved problems of AI. No one ever responded. I would be interested in hearing of any systems you uncover. I really don't see the need for NL understanding in a parser for such a limited domain; a fairly simple expert system should be able to do the job. Even a YACC-style parser might be sufficient, especially if boldface and italic can be discerned. If an online citation database is available, one effective (if inelegant) approach would be to treat the human-readable citation as a set of keywords for retrieving more complete or correct citations in the formatted database. -- Ken Laws ------------------------------ Date: Wed, 25 May 88 14:39 EDT From: rwojcik@BOEING.COM Subject: What are grammars (for)? Arild Hestvik writes: AH> The fact that you can create a context where the Binding Theory doesn't AH> work doesn't mean that the Binding Theory is wrong, it simply means that AH> you set up a lousy experiment which yielded garbage as a result. That is AH> very easy, you can do that to everything you learnt in high-school physics, AH> simply by doing the experiment wrongly. It is very difficult to set up experiments to confirm or disconfirm anything in generative theory. Postal's satire, "Advances in Linguistic Rhetoric" (NLLT Feb. 1988), is the best explanation I have seen of why that is the case. I did not set up a context where "Binding Theory doesn't work." I merely showed the futility of trying to build a theory of language understanding on a syntactic theory that purports to work independently of pragmatic context. You can't take a sentence such as 'He shaves John' and claim that the NPs necessarily refer to different individuals. Nor can you claim that 'John shaved himself' is necessarily a case of two NPs that refer to the same individual. Note that you can say 'John went back in time and shaved himself', where 'himself' could refer to the original or a past version of John. But 'John went back in time and shaved' unambiguously means that John shaved his original self. The semantics of lexical and syntactic structures behaves differently with respect to referential analysis. ('X shaved himself' means 'X shaved Y and X=Y', whereas reflexive 'X shaved' means 'X shaved X'. The question of what validates 'X=Y' is nontrivial, to say the least. But it is a question involving context.) AH> ... a theoretical linguist believes that there is such a thing AH> as knowledge of language that is independent of the use of it... Not all theoretical linguists believe in the competence/performance dichotomy. If you reject the dichotomy, then you are not a *generative* theoretical linguist. It would be interesting to know how many theoretical linguists truly believe in, only pay lip service to, or outright reject the dichotomy. (Feed that to your parser and weep! :-) I would say that the majority fall into the middle group. They consider themselves generativists, but they don't attend church too often. ------------------------------ Date: Wed, 25 May 88 17:12 EDT From: Richard A. O'Keefe <quintus!ok@Sun.COM> Subject: Re: 3rd person pronouns > In article <5614@bcsaic.UUCP> rwojcik@bcsaic.UUCP (Rick Wojcik) writes: > > What we could really use is a truly > > genderless singular 3rd person pronoun such as that found in Finno-Ugric > > languages (Hungarian, Finnish, Estonian). We could borrow the word 'ta' > > (he/she/it), for example. How about it? I'll use ta if everyone else > > will. So now we say "Everyone flipped ta lid" instead of the sexist > > "Everyone flipped his lid." [...] There is what appears to be a false presupposition in that, namely that pronouns-with-gender are "sexist". Has anyone actually done a cross- cultural study that demonstrates a correlation between having or not having gender-marked pronouns and social gender attitudes? The only non-European languages I have any knowledge of are Pijin and Maaori, neither of which has gender-marked pronouns and neither of which is associated with a sexually equalitarian culture. The word "ti" has been used; I have in my possession an MSc thesis written in England which uses "ti" throughout. This appears to be a coincidence, "ti" is "it" spelled backward. But instead of faking something up or borrowing it from an unrelated language, why not either (a) extend existing English usage or (b) adopt a word from a closely related language? (a) Some British dialects already use "it" for babies and young children. Since this is comp.<<AI>>.nlang-know-rep, would it not be as well to eliminate "animatism" from the language as well as "sexism" (:-)? (b) Pijin has "mi" (pronounce "me"), "u" (pronounce "you"), and "i" (pronounce "ee") as singular pronouns. They do not inflect. Adopting "i" from Pijin (spelling it "e") would give us nom: "e", acc: "e", gen: "e's", ref: "eself". "Ta" would be a bad idea: it is already an (informal) English word. More homonyms we can do without. Best of all, avoid false consciousness, try to change realities, not labels. ------------------------------ Date: Wed, 25 May 88 18:25 EDT From: Miriam Eldridge <miriam@olivej.olivetti.com> Subject: Re: genderless 3rd person singular pronoun The Finnish genderless 3rd person singular pronoun is "han" (pronounced with a "short a" and properly spelled with an umlaut, but Unix knoweth not diacritics nor the IPA). I suspect this was borrowed from Swedish. However, its associated nightmarish case system is purely Finnish: gen. hanen, acc. hanet, dat. hanelle, etc., etc., up to 15 cases. Miriam Eldridge (Suomalainen) ------------------------------ Date: Thu, 26 May 88 12:40 EDT From: Robert France <france@vtopus.cs.vt.edu> Subject: Re: Genderless 3rd person pronoun. In regard to the discussion by Rick Wojcik and Rich Alpert (nl-kr v4 #53), those seeking an excellent example of the use of a genderless pronoun are referred to *The Cook and the Carpenter*. A feminist novel from the early seventies, it was written entirely using the pronouns "te", "ter", and "ters" and using people's titles instead of names. The result is an interesting cognitive dissociation in the reader's mind: you're not sure of the sex of ANY fo the main characters until midway through the novel. It's also a reasonably good read -- exciting, neat characters, and so forth. I'm afraid I don't have the author's name easily to hand. On a more pragmatic note, I've been noticing that "they" and "them" are entering spoken English as genderless singular pronouns. In written English, though, I still prefer "s/he", "hi/r", and "hi/rs". "Their minds are confused with confusion, And to their problem, their is no solution." -- Robert France france@vtopus.cs.vt.edu Department of Computer Science Virginia Tech Blacksburg, VA 24061 ------------------------------ Date: Thu, 26 May 88 20:06 EDT From: Adam J. Kucznetsov <adam@cunixc.columbia.edu> Subject: Re: genderless 3rd person singular pronoun In article <4645@husc6.harvard.edu> alpert@endor.UUCP (Rich Alpert) writes about a suggestion to use 'ta' as a genderless 3p.s. pronoun: >What about cases? "Everyone flipped ta's lid."?? "Give it to tam."?? >(pronounced "Tom"?) Consider: "Someone is at the door. See who it is and ask >ta what ta wants." ^^ Someone is at the door. See who it is and ask tam [what ta wants | tas name -- ta looks cute! ] My own suggestion for a genderless pronoun is: 3s 3p subj. se they obj. hem them gen. hes their The stressed pronunciation that I use is "se" as "she," "hem" and "hes" as "him" and "his" with the vowel souund as in the existing word "hem." Absolutely no etymological basis for this series -- I simply found it easy to adopt. It seems to my ear somehow more natural than "ta, tam, ta[']s" even with its synthetic irregularity. I arrived at it, of course, by combining elements of the existing "he, ..." and "she, ..." he + she => se him + her => hem his + her => hes There need be no arguments concerning the degree to which "se, ..." borrows from the masculine or the feminine pronouns. :-) Both "hem" and "hes" can be said to take two letters from either the masculine or the feminine and the third from the other. "Se" obviously takes one from each, -- or is derived from "he" by replacement or from "she" by deletion of the formerly masculine "h". Someone is at the door. See who it is and ask hem [what se wants | hes name -- se looks cute!] Everyone flipped hes lid. If someone wants the tape that badly, [give it to hem | give hem it.] Almost every senator thinks se is above the law. I often use this series, both in conversation and writing, when it's not too confusing for the listener or the speaker. I always use it when se already knows about my peculiar convention. cat -- Adam (Cat) J. Kucznetsov| adam@columbia.edu ui.adam@cu20b (BITNET) "He saved the world and | adam@cunixc.columbia.edu ajuus@cuvma (BITNET) called it art..." | cat@cubsun.bio.columbia.edu adam@cs.columbia.edu + UUCP: {rutgers, topaz}!columbia!adam ------------------------------ Date: Wed, 25 May 88 19:24 EDT From: Jim Meritt <jwm@stdc.jhuapl.edu> Subject: Benjamin Lee Whorf I have noticed that people in this newsgroup seem very down on the S-W hypothesis. What information (not someone's opinion, I grow my own) exists pro/con this hypothesis? If the time is not available to post/email the data, could you do the same with the references? The library here is very heavy on physics, and almost all the language references concern non-human languages, but they will inter-library loan them for me IF I have a title, author, date,... (usual id stuff). Thank you. p.s. I have _Language, Thought, and Reality_ on hand, so no need to give that as a ref, though you might want to point out particular items of note. Disclaimer: Individuals have opinions, organizations have policy. Therefore, these opinions are mine and not any organizations! Q.E.D. jwm@aplvax.jhuapl.edu 128.244.65.5 (James W. Meritt) ------------------------------ Date: Tue, 24 May 88 10:11 EDT From: Gordon Fitch <gcf@actnyc.UUCP> Subject: Re: Observation ("Back to the Future"?) In article <657@taux01.UUCP> amos@taux01.UUCP (Amos Shapir) writes: >See I Samuel 9::9 for a Biblical counter-example, i.e. reference >to the past as being in front (actually, 'before' has the same meaning >in English). The past is thought of as being _before_ other events when all are considered to be at a distance -- as if the events were on parade. Thus in "Macbeth", when Macbeth has a vision of the descendants of Banquo, he sees them as passing in single file with the oldest first. This is a different frame of reference than when the speaker himself becomes involved, and what is ahead is considered to be "ahead" as if the speaker were walking forward -- in English, anyway. My handy version of the Bible is the KJV, and I don't know what the Hebrew usage is. Then there's "My whole life passed before my eyes." All of which does imply that the use of spatial terms for describing temporal relationships is largely conventional. In spite of the fact that we all know very well that the future is harder to see than the past, we who speak English don't say the future is behind us, although others may. It would be interesting to find out if speakers of languages with differing conventions had differing senses of the unknowability of the future, and so on. ------------------------------ Date: Wed, 25 May 88 21:09 EDT From: doug-merritt@cup.portal.com Subject: Re: Observation ("Back to the Future"?) > It would be interesting to find out if >speakers of languages with differing conventions had differing senses >of the unknowability of the future, and so on. For a summary of research on even more general issues than this, see "Cross-Cultural Universals of Affective Meaning", Osgood/May/Miron, Univ. of Illinois Press, copyright 1975. It doesn't answer your particular question, but it sure examines such issues in general. Does anyone know of a followup work to this classic? Doug -- Doug Merritt ucbvax!sun.com!cup.portal.com!doug-merritt or ucbvax!eris!doug (doug@eris.berkeley.edu) or ucbvax!unisoft!certes!doug ------------------------------ End of NL-KR Digest *******************