5-1362msg1.txt

运用贝叶斯网编的一个识别垃圾邮件的程序

Subject: number invariance : summary

several week ago i post a call for help identify language in which noun may show the same form of number with all enumerate numeral . example from english , which do not in general allow such number invariance , include the noun ' head ' in ' one head of cattle ' , ' two head of cattle ' , etc . and , more systematically , noun compound with numeral ( cf . two-car garage , three-body problem , four-week in - titute , five-page summary , six-foot high ) . a summary of the response follow . please feel free to add to this list ( or to take issue with my summary ) . thank to patrick farrell , tim pulju and all the source mention below . thank also to the re - pondent who provide the follow general discussion of number invar - iance and english compound : churma , d . ( 1983 ) " jet fan , raider rooter , and the interaction of morphosyntactic process " . cls 19 ( paravolume ) churma , d . ( 1987 ) " explain level order , and how not to parse a word " . bls 13 . rijkhoff , j . 1991 . " nominal aspect " . journal of semantic 8 - 4 , 291-309 . rijkhoff , j . 1992 . the noun phrase : a typological study of its form and structure . doctoral dissertation , university of amsterdam . [ esp . ch . 3 . 1 . 1 . pp . 74-103 ] rijkhoff , j . forthcome ? . " ` number ' disagreement " . proceeding of the xvth international congress of linguist , 9-14 august 1992 , quebec , canada . 1 ) australian : many australian language show number invariance . although they generally have a plural morpheme that may be affix to noun , this be not use unless require by pragmatic considera - tion . thus , in pitjantjatjara ( western desert ) , ngampu be ' egg ' , and we have ngampu marnkurpa ' three egg ' . reference on pitjantja - tjara : goddard , cliff . 1993 . a learner 's guide to pitjantjatjara / yankunytjatjara . alouse spring : institute for aboriginal development trudinger ; ronald m . 1943 . grammar of the pitjantjatjara dialect , central australium . oceanium xiii : 3 , 205-223 source : rob pensalfinus ( rjpensal @ mit . edu ) 2 ) basque : noun in indefinite nps behave like possibly all turkic language in use an invariant form of the noun with all numeral . example : gizon bat ` one man ' , bus gizon ` two man ' ( bizkaian gizon bus ) , hiru gizon ` three man ' . noun plurality may be mark , however , in nps contain definite determiner . one such determiner be the " ordinary " or " definite " article - a , which be a suffix . example : gizon ba ` the one man ' ( rare ) , bus gizonak ` the two man ' , ` both man ' ( bizkaian gizon biak ) , hiru gizonak ` the three man ' , ` all three man ' . reference : la - fitte , pierre ( 1944 ) , grammaire basque , pp . 76-78 . saltarellus , mario ( 1988 ) basque , p . 172 . source : larry trask ( larryt @ cog . susx . ac . uk ) 3 ) celtic : in welsh ( an perhap in all celtic language ) , all cardinal number obligatorily take thesingular of the follow - ing noun ( cf . cus 'd og ' , cwn 'd og ' : un cus , dau gus , trus chus , pedwar cus , pum cus , chwe chus , saith cus , wyth cus , naw cus , deg cus . . . ( ' one dog , two dog . . . ten dog ' ) . there be , however , an alternative partitive construction , especially favour with higher number , and here you get the plural : pump o gwn , chwech o gwn . . . ( light . ' five of dog , six of dog ' ) . any welsh grammar book will give you a reliable description . reference : ball , m . ( ed ) ' the celtic language ' routledge , 1993 . source : nigel love ( nlove @ beattie . uct . ac . za ) martin j . ball ( mj . ball @ ulst . ac . uk ) 4 ) chaha : in chaha there be no plural morpheme ( cf . at bet ' one house ' , xwet bet ' two house ' ) , except in some lexically mark plural on a very small number of noun ( cf . at arc ' one boy ' , xwet dengya ' two boy ' ) . source : banksira degif petro ( d217704 @ er . uqam . ca ) 5 ) chinese : chinese do just what you write : yus tou niu ' one tou cow ' , where tou be a classifier for domestic animal ( and also a noun in its own right mean head ) . source : paul wood ( woodspr @ osuunx . ucc . okstate . edu ) 6 ) choctaw : choctaw noun be never mark for number under any cir - cumstance . number mark be diffuse and lexical , mean there be a variety of non-parallel strategy for indicate the number value of particular noun . source : marcium haag ( haag @ monk . nhn . uoknor . edu ) 7 ) dutch : dutch word for time interval like kwartier ' quarter ' , and uur ' hour ' be alway singular . maand ' month ' and jaar ' year ' usually remain singular . seconde , minuut , dag 'd ay ' , nacht ' night ' , week , and eeuw ' century ' take plural morphology when there be two or more of them . meter , centimeter , kilometer , hectoliter , kilo - ( gram ) and frank , alway remain singular too . source : bert peeter ( peeter @ postoffice . uta . edu . au ) patricium haegeman ( fte . haegeman . p @ alpha . ufsium . ac . be ) 8 ) english creole of new guinea ( = neomelanesian ? ) : here we have example like wan pelum man ' one man ' ( light . ' one fellow man ' ) , tu pelum man ' two man ' ( light . two fellow man ) . similar fact may be find in ( some ) of the underlie local langua - ge . source : brian drayton ( brian _ drayton @ terc . edu ) from personal communication with anthony arlotto , e . wayle browne ( ewb2 @ cornell . edu ) . 9 ) finno - ugric : hungarian have a plural morpheme - k . yet , in spite of plural like toll ( a ) k ' pencil ' , one say egy toll ' one pencil , ket toll ' two pencil ' , harom toll ' three pencil ' . . . , sok toll ' many pen - cil ' . further example : kapu ' gate ' have a plural kapuk ' gate ' . yet we have hat kapu 's ix gate ' . finnish work the same way . source : roman agne ( aromus @ erato . erin . utoronto . ca ) edith a moravcsik ( edith @ csd . uwm . edu ) 10 ) georgian : generally , noun follow a cardinal number remain in the singular . measure word be not use . source : john peterson ( gor05 @ rz . uni-kiel . d400 . de ) 11 ) german : gerhard helbig & joachim buscha , deutsche grammatik say on p . 528 that some measure noun ( mostly feminine ) di - tinguish between singular and plural in phrase like eine fla - che sekt " a bottle of sparkle wine " - zweus flaschen sekt . a list follow : die buchse , die dose , die kiste , der krug , die schussel , der tag , die tasse , die woche . . . other measure noun ( mostly neuter ) keep the singular : ein stuck zucker - dreus stuck zucker , ein kasten bier - dreus kasten ( or dreus ka " sten [ the pl ] ) bier . likewise : da blatt , da gla , da kilo , die mark , da meter , da paar , da pfund , der sack . . . ( the . . . be in the original ) . soruce : e . wayle brown ( ewb2 @ cornell . edu ) 12 ) indic : in hindus , a measure of time , distance , mass , etc . be leave in the singular with cardinal number number . otherwise , common noun appear in the plural after a number . measure word be not use . as with hindus , a measure of time , distance , mass , etc . be leave in the singular in nepalus . in addition , common noun here be generally leave in the singular and one of two measure word be use : janaa for human , vataa for animal and thing . this un - doubtedly be due to tibeto - burman influence . source : john peterson ( gor05 @ rz . uni-kiel . d400 . de ) 13 ) indonesian ( bahasa indonesium ) : although one may pluralise a noun by reduplication , this be a non-specific form of the plur - al ( cf . kuce ' cat ' , kucing-kuce ' cat ' ) . if one want to say five cat , it would simply be : lima ekor kuce ' five classifier cat ' . i be pretty sure that ' lima ekor kucing-kuce ' would get a star , but i ' ll give it a provisional question mark . reference : mac - donald , r . ross . 1976 indonesian reference grammar . washington , dc : georgetown university press . source : rob pensalfinus ( rjpensal @ mit . edu ) 14 ) japanese : have no inflectional morphology for noun , and thus ex - hibit number invariance completely . source : steve seegmiller ( seegmiller @ apollo . montclair . edu ) 15 ) kiribatus ( = gilbertese ) : the unmark form bentira ' pencil ' be use for example in a . aus bentira ' four pencil ' ( light . ' four . long unit pencil ' ) and a . botakus bentira ' four group of pencil ' ( light . ' four . group pencil ' ) . source : martin silverman ( mg @ unixg . ubc . ca ) 16 ) modern persian : in an afghan variety of persian one can say yak bacha ` one brother ' , do bacha ` two brother ' , se bacha ` three broth - er ' , char bacha ` four brother ' , etc . reference : any grammar of modern persian . also carleton hodge have numerous publication descrus - bing aspect of persian structure . source : charle scott ( cscott @ macc . wisc . edu ) 17 ) newarus : this tibeto - burman function like nepalus ( see indic above ) in this respect . the main difference be that newarus almost alway have a measure word after a cardinal number . generally , only noun denote live being have a plural form . reference : han joergensen , a grammar of the classical newarus . copenhagen : ejnar munksgaard , 1941 . ( det kgl . danske videnskabern selskab , hist . - filol . meddelelser , xxvii , 3 . ) source : john peterson ( gor05 @ rz . uni-kiel . d400 . de ) 18 ) niger - congo : for bambara , c . bailleul ( cours pratique de bambara , iii : type de phrase . imprimerie de la savane . bobo - dioulasso , 1977 ) only mention cn num , while c . bird et al . ( an ka bamanan kalan : begin bambara . iulc , 1977 ) say an addition - al possibility be cn num pl - def ( p . 43 ) . dogon seem to have just cn num ( cf . kervran , m . & a . prost le parler dogon i . donno so . document linguistique 16 , universite de dakar , 1969 ) . source : chri culy ( cculy @ vaxa . weeg . uiowa . edu ) 19 ) semitic : semitic often make " no statement as to number " . noun have regular plural , but in count you use the plural form only up to 10 and then revert to the singular . in hebrew this be most normal with time word , which happen also to be a small group of word that have a dual , as well . so for " yom " ( day ) you count " yom exad " , yomayim ( the dual ) , and from 3 to 10 it 's " shlosha yamim . . . asara yamim . " then you usually switch back to " axad-asar yom " ( 11 ) , etc . standard arabic count be far messier with case interfere , too , though all modern speak arabic be more like hebrew . ( they do n't have a syntax that would allow complression into " a three-day conference " type of phrase . ) source : bob fradkin ( raf100f @ oduvm . cc . odu . edu ) 20 ) siouan : siouan language mark number in the verb , not the noun . noun , so there be no plural morphology for noun , they be natur - ally invariant with numeral ( numeral be a sort of verb , actually some of the language can mark number in nps by choose among var - ious form of the definite article , which mark positional / configur - ational gender . collection of thing have a different configura - tion from single item , so a different article . source : john e . koontz ( koontz @ alpha . bldr . nist . gov ) 21 ) turkish : turkish and perhap all turkic language have a plur - al inflection - ler / - lar which be usually omit if a numeral mark the np as plural ( unless definiteness be express ) . example : adam ` man ' , adamlar ` man ' , ` the man ' , bir adam ` one man ' , ikus adam ` two man ' , ikus adamlar ` the two man ' , ` both man ' . reference : lewi , tur - kish grammar , clarendon press , pp . 25-26 , or underhill , turkish gram - mar , mit press . source : edith a moravcsik ( edith @ csd . uwm . edu ) steve seegmiller ( seegmiller @ apollo . montclair . edu ) larry trask ( larryt @ cog . susx . ac . uk ) e . wayle browne ( ewb2 @ cornell . edu )