5-1285msg3.txt

贝叶斯学习算法源码

Subject: language evolution context

> from : gerard gautier < gauier @ cc . nsysu . edu . tw > [ interest stuff ] > i sometime realise that , while talk to my sister in france by > phone , i find myself unthinkingly tell her : > > " j ' habite dan une petite hsiang derriere l ' ecole . " ^ ^ ^ ^ ^ ^ ditto here , with english . need i describe the puzzle look when once i let " le achevement academique " slip into my french ( " cesar achevant le prisonnier gauloi " , par salvador dalus , a la maniere de meissonnier ! ) . the case of " hsiang " be different , though , just as that of " le milk-bar du coin " ( which my wife and i constantly use ) . i expect that " ruelle " be not strongly associate in one 's experience with the thing , but it would be if you have be live in a french - speak chinese town . since it be usually not french but chinese which you hear in chinese town ( mercus , monsieur de la palisse ) , " hsiang " be strongly associate , so much so that it creep into even a french sentence . ditto with " milk-bar " in australium , and , i guess , " trinkhalle " in germany . the closest i can think of in french would be " l ' arabe du coin " , but that would be only parisian french i imagine . > so do it not mean that the evolution should depend on the education > level of people - so that it would be differential inside a society > accord to , say , the social class ? well , great britain provide prime example of cant . i have come across , a few week ago , " language in the british isle " , peter trudgill ( ed ) , cambridge university press , 1984 , isbn 0-521 - 28409 - 0 . i cannot recommend it too highly . chock - a-block full of engross , well-present datum , from " rural dialect in england " to " romanus and angloromanus " vium " shelta and polarus " . even discount cant , i remember my father mention have start learn breton ( we live in douarnenez , then with breton speaker ) , and give up after find the breton he learn from the fisherment be well nigh useless for communicate with the farmer . back to cant , be n't francoi villon 's poem in argot incomprehensible to-day ? > - i be also aware of phenomenon of resistance to acculturative evolution > ( conscious or not ) as for the quebecquoi tell , not " un hot-dog " as > the french do , but " un chien chaud " ( toubon be in my opinion but a > marginal impedement on the natural evolution , whatever direction it take ) . . . > > - another exemple of the complexity of those issue would be the > " rigidification " of a language face a threat . > ( phonology , vocabulary , syntax . . . ) . two facet of the phenomenon , be n't it ? > to resume my question : how can we ascertain that this language > have have this reaction without have to resort to extra-linguistic > implicit hypothesis as : homogeneity through class , identity of > evolution whatever the number of speaker would be ( lng-switch v > creolisation . . . ) , existence ( or not ) , type , and evolution of a script > system ( * * ) etc . . . ? can we ? and further , can we , even resort to extra-linguistic hypothesis ? those hypothesis seem to me post hoc , and hold only for a subset of the datum . in other word , you may find counter-evidence galore if you do not blind yourself to it . further , have those hypothesis any useful predictive value ? what i mean be this : they may be significant at 99 . 99 % confidence level and beyond , yet remain useless . example : have record the outcome of 100 , 0 spin at the roulette table , you find 50 . 5 % red and 49 . 5 % black , grab you pocket calculator , and yes , yes , yes , that particular roulette wheel must be bias toward red . with dream of untold riches prod you , you start bet on red . you may realize now that you after all you be not go to get very rich very quickly with just a 0 . 5 % edge : your ability to predict the outcome of the next spin be , for all practical purpose , unchanged . not only that but , unless you have almost as much money to play with as the casino , you be certain to be wipe out in the long run . but that be another story , call " the drunkard 's walk " .