5-1254msg1.txt

运用贝叶斯网编的一个识别垃圾邮件的程序

Subject: typological classification

a few remark on fritz newmeyer 's post on classification in language typology : of course it be not alway easy to decide to which type a give language should be assign , since there may be conflict criterion . but some people seem to think that this be somehow a deep problem with language typology . in reality , there be no deep problem here at all , so the typological literature be quite right in not dwell on such trivial methodological point too much , concentrate on substantive and theoretical issue instead . true , the methodology of language typology be somewhat different from the methodology use in single-language study that most linguist engange in . when search for significant correlation between feature / parameter in the world 's language , one have to classify language , and the more language one look at , the less energy one can spend on resolve each individual case . clearly , the speacialist will sometime disagree with the way her or his language be assign , but then most of them time not all specialist agree on the right classification themselve . when the criterion be in conflict , it may sometime be necessary to give more weight to some than to other in a somewhat arbitrary way - - but this be simply one reflection of the necessary idealization that accompany any serious scientific endeavor . it 's as simple as that : the more one study linguistic phenomenon in breadth , the more one lose in depth . but vice versa , linguist that study only one or a few language be work on an extremely narrow basis and lose in breadth what they gain in depth . ideally , typological research should be conduct by team of linguist , much in the same way as research in big science be organize . with more resource that be combine , one could study linguistic phenomenon in considerable breadth and depth simultaneously . an attempt to follow this strategy have be make in the european science programme in language typology , which have be run in europe for the past five year ( wait for 10 volumes on the typology of european language , to appear with mouton de gruyter in a year 's time ) . this be only a very modest , seriously underfund attempt , but until we attract really big research money or manage to agree on more thing , we have to be happy that we have colleague like johanna nichol who be will to make enormous effort to get a view of linguistic phenomenon on a global scale . martin haspelmath ( free university of berlin )