5-1420msg2.txt

运用贝叶斯网编的一个识别垃圾邮件的程序

Subject: re : 5 . 1404 comparative method in linguistics

karl teeter be mistake , i think , when he say that you cannot classify language on the basis of phonological correspondence in the lexical item , when he say that you would treat english and french as relate ( vium the norman borrowing ) if you do this , when he say that the idea of use this method be " novel " , when he say that you can only classify lamnguage as relate if you have write a comparative grammar , etc . ( a ) for many language family , there is no other basis for classification available , because they lack the kind of morphological complexity so belove of indo - europeanist , algonkianist , and those other - ist who happen to enjoy its blessing in " their " language , ( b ) the danger of confuse borrowing for cognate be alway real , but it be easy to see that the lexical connaection of english with french be more transparent phonologically and less central semantically than those with dutch , hence more recent , hence borrowing . ( c ) there be nothing novel in what i be say , since it be the method which , for example , edward sapir use to establish that the uto - aztecan language be really a family ( rather than three family ) . ( d ) you cannot possibly realistically expect normal people to spend time write comparative grammar of language which have not previously be show to be relate . so classification must come first ! on the other hand , it be perfectly easy to write a " crazy " comparative grammar for any random group of language , e . g . , french and english . the last be perhap the most important point : reconstruction of morphological system can be do well or badly , so can comparison of lexical item . both can yield correct result ; both can also produce garbage . p . s . i say ' lexical item ' deliberately , because there seem to be language family where there be no morphology at all to compare . in most case , howveer , there be some bind morpheme , and one compare those as well ( as do sapir in the case of uto - aztecan ) . but this be a far cry from reconstruct the kind of intricate morphological pattern find in the older indo - european language , for example , but absent from many other language family .