5-1393msg1.txt

运用贝叶斯网编的一个识别垃圾邮件的程序

Subject: comparative method in syntax

fritz newmeyer ask if it be legitimate to apply the comparative method in syntax . my answer be ye , but it be more difficult . the main reason why one see more work on historical-comparative syntax than fifty year ago be that syntax have become much more prestigious in recent decade . i do n't think that there be a shift from a once " standard " view that syntactic reconstruction be impossible . bernhard delbrueck 's historical-comparative syntax of indo - european , publish a hundr year ago , clearly show that comparative syntax be both possible and fruitful , but perhap too difficult to attract many follower . the view that historical-comparative syntax be impossible be defend in some detail in lightfoot 's ( 1979 ) principle of diachronic syntax , but few researcher seem to have be discourage by lightfoot 's attitude . it be true that latin syntax could hardly be reconstruct from modern romance language , but neither could latin morphology , and even the view of latin phonology that we would get from romance be very distort . our reconstruction of protolanguage grammar be alway imperfect , but that be no reason not to attempt reconstruction . the main error on lightfoot 's side be that there be no syntactic analog to the regularity of sound change . it 's just a bite more complicate in syntax - - just like syntax be more complicate overall than phonology . basically , the analog to phoneme and word be word and sentence . true , the difference between word and sentence be that in general sentence be not store in the lexicon . but neither be many complex word ( render morphological reconstruction similarly difficult , but nobody seem ever to have object to morphological reconstruction ) , and some sentence be actually store , e . g . proverb and idiom , which often show syntactic archaism . and since earlier syntax often survive in " fossilize " form in later morphology , we have another rich source of datum for diachronic syntax . in addition to regularity of change , we need general principle of change for plausible reconstruction , e . g . phonological principle that predict likely change like assimilation , lenition , segment loss , etc . in syntax , similar principle of change exist as well : spatial noun become spatial adposition , certain general verb become tense and aspect marker , allative case marker become dative case marker , purposive verb form become infinitive , etc . all these process ( instance of grammaticalization ) be irreversible change and provide safe guide for linguist seek to make sense of daughter language diversity by reconstruct a proto-syntax . the massive regularity of grammaticalization be generally ignore in generative study of syntactic change ( indeed , lightfoot argue that there be no genuine principle of diachronic syntax ) , but if one take them into account , they help in the difficult task of reconstruction . martin haspelmath ( free university of berlin )