资源描述：
《Unit 6 Making statistical claims》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。
1、Unit6Makingstatisticalclaims6.1IntroductionOneofthemostobviousadvantagesofusingacorpus,ascomparedwithintuition,isthatacorpuscanprovidereliablequantitativedata(cf.unit1.5).Inthisunitwewillconsiderwhattodowiththequantitativedatathatcorporaprovide.Ourapproachtothistopicwillberealistic–byandlarge
2、mostusersofcorpusdatawillnotbecapableofgeneratingsophisticatedstatisticalclaimsnorwouldtheywishtodoso.However,wewantreaderstobeawareofwhattheyshouldnotclaimonthebasisofthesimpledescriptivestatisticsthattheymayuse.Also,inthisunit,wewanttogiveguidanceonhowtointerpretinferentialstatisticsgenerat
3、edbytheconcordanceprogramsusedinSectionC.Whilestatisticscanbeintimidatingformanyreaders,abasicawarenessofstatisticsisessentialwhenadoptingacorpus-basedapproachas‘theuseofquantificationincorpuslinguisticstypicallygoeswellbeyondsimplecounting’(McEneryandWilson2001:81).Thisunitwillcoverthebasics
4、tatisticalconceptsandtechniquesrequiredtounderstandsomeexcerptsinSectionBandthecasestudiespresentedSectionCofthisbookwhiletryingtoavoidacomplex,technicaltreatmentofstatistics.Someoftheseconceptsandtechniques,though,willbefurtherdiscussedinthecasestudiespresentedinSectionCofthisbook.6.2Rawfreq
5、uencyandnormalizedfrequencyIncorpuslinguisticsfrequencyreferstothearithmeticcountofthenumberoflinguisticelements(i.e.tokens)withinacorpusthatbelongtoeachclassification(i.e.type)withinaparticularclassificationscheme(e.g.theCLAWStagset,seeunit4.4.1).Itisthemostdirectquantitativedataacorpuscanpr
6、ovide.Typically,frequencyitselfdoesnottellyoumuchintermsofthevalidityofahypothesis.Yetdataofthistypecanbeusedinbothdescriptivestatisticsandinferentialstatistics(seeunit6.3).Note,however,thatfrequencydatamustbeinterpretedwithcaution.Itispossibletouserawfrequency(i.e.theactualcount)wherenocompa
7、risonbetweencorporaisnecessary.However,whencomparingcorpora(orsegmentsinthesamecorpus)ofmarkedlydifferentsizes,rawfrequenciesextractedfromthosecorporaoftenneedtobenormalizedtoacommonbase(seecasestudies2,4,5and6).Forexample,theswearwordfucker(
