Limits On Phonetic Accuracy In Foreign Language Speech .

1y ago
7 Views
2 Downloads
1.52 MB
14 Pages
Last View : 22d ago
Last Download : 2m ago
Upload by : Nora Drum
Transcription

Limits on phonetic accuracy in foreign language speechproductionJames Emil sbarns35209JamesHillenbrandDepartmentof CommunicativeDisorder llinois32601(Received20 June1983;acceptedfor publication27 April "}and/ty/("tu"} producedin handFrenchtalkerslivingin theU.S. In ,andmorevowelsproducedby sproduced/u/withsignificantlyhigherF2 valuesthantheFrenchtalkers,butproduced/y/withF2 valuesequaltothoseof the Frenchtalkers.A labelingtaskrevealedthat the/y/.vowels producedby ceda moreidentifiable/u/than the inexperiencedAmericans.It idly becausethisvowelisnot--likeFrench/u/--judged to beequivalentto a vowelof h equalVOT valuesof about55ms,whichisintermediateto valuescommonlyobservedfor monolingualspeakersof FrenchandEnglish.It ishypothesizedthat thebilingualtalkersjudgedthe/t/of FrenchandEnglishto be equivalent,whichaffectedtheirperceptualtargetfor French/t/and ultimatelytheirproductionof thisstop.PACS numbers:43.70.Ve, 43.70.Dn, 43.70.Gr, 43.70.BkINTRODUCTIONMost adultswholearna foreignlanguagespeakwith an"accent"whichderivesin part fromphonologicaland phoneticdifferencesbetweentheirnativelanguage(LI) andthetargetforeignlanguage(L2}.This raisesthe questionof theextentto whichhumansare capableof learningnew speechpatterns,or modifyingexistingones,onceL1 hasbeenthoroughlyacquired. This study examinedfactorsthat mightlimit theaccuracywith undin a quentlyobservedthat L2 learnerstendto producethe wordsof a foreignlanguagewith phonesthatcanbeheardin thephoneticsurfaceofLl. "interlingualidentification"of L1 andL2 phones.Valdman {1976,p. 38) observedthat:hapsarticulatory,similarityofLI andL2phones.However,empiricalresearchhas yet to: {1) objectivelyquantifythedegreeof phoneticsimilaritybetweenphonesfoundin twolanguages,(2)demonstratewhichspecificphonesin LI andL2 are "identified" with one another, or (3) determinewhetherthe interlingualidentificationof Ll and L2 phoneswill ceaseas'thelearnerbecomesfamiliarwith thephoneticdifferencesthat may distinguishLI and L2 phones.SomeL2 phoneshavea directcounterpartin L1 withwhichthey can be identified.To a first approximation,thisincludesLI and L2 phonesthat are transcribedusingthesameIPA symbol.For example,instancesof/t/occurringin FrenchandEnglishwordsarelikelyto beregardedby theL2 learneras beingdifferentrealizationsof the samecategorybecauseof their overallphoneticsimilarity.The interlingualidentificationof suchpairsmightbe expectedto occur evenwhenacousticdifferencesthat may distinguishtheL I and L2 phonesare icallydifferentphonesto bemembersof the samecategoryis a fundamentalaspectof human.thearticulatoryhabitsof theforeignlanguagepartially overlapthoseof the nativelanguage. The studentmustlearnto makenewresponsesto stimuliwhichareinterpretedasidenticalto ,French and English/s/differ with regardtoOther L2 phonesbear less obviousresemblancetoplaceof articulation.The formeris a dental,and thephonesin LI andmaythereforenot be regardedasthe ceives izationof any LI category.For nativeEnglishspeakers,thisin theacousticsignalof French/s/will leadanEnglish may includephonessuchasthe front-roundedvowel/y/ofspeakerto respondwith the alveolarrather than theFrench, the clicks of SouthernBantu languages,and thedental sound.pharyngealfricativesof Semiticlanguages.We will refer tosuchphonesas "new," althoughit shouldbe apparentthatThis representsthehypothesisthat whenan L2 phoneisany phoneencounteredin a fo ign language---nomatter"identified"withanL! phone,theLl.phonewill beusedinplace of it {i.e., be "substituted"for it). Such interlingual howexotic-- islikelyto o phonesfoundin the learner'sLI.identificationappearsto dependon the auditory,and per708J. Acoust.Soc.Am.76 (3), September1984 0001-4966/84/090708-14500.80 1984 AcousticalSocietyof America.708

One aim of this study was to determinewhetherL2learnersproduce' ew" L2 rtinLI. Valdman(1976)hypothesizedthatnewL2 phonesarelearnedmoreeasilythanL2 phoneswhiohhaveanLI ionandthustendnot to beproducedwith previouslyestablishedpatternsof segmentalarticulation (cf. Koutsoudasand Koutsoudas,1962).Koo (1972)suggestedthat talkersdo not actuallyneedto "learn" newL2 phonessincetheycanbe producedby recombiningthefeaturesof L1 phones.ThisimpliesthatnewL2 phonesandthosewhichhavea directcounterpartin LI will beproducedwith 1969; ohansson,1973;Walz, 1979;liege and Port, 1981),but nonehasspecificallytestedthe hypothesisthat new L2 phonesare producedmoreaccuratelythanthosewith a directcounterpartiencedL2 learners(seeliege, 1984b).A nonauthenticpro-nunciationof vowelsleadsto foreignaccent(E!sendoorn,1983)butnostudy,to ourknowledge,hasspecifi y examinedtheeffectof experienceonlearners'productionof L2vowels.Thus anotheraim of this studywas to determinewhetherexperiencedAmericanspeakersof FrenchproduceFrench vowels more accuratelythan less experiencedAmericansand,if nd/u/.Most previousL2 researchhasemphasizedthe difficultyof establishingnewmotorplansfor L2 phones,or thedifficultyin modifyingpre*existingones.The final aim ofthisstudywasto testthehypothesisthatanimportantcauseof foreignaccentis the developmentby L2 learnersof ,1984b)hypothesizedthat interlingualidentificationleadsthe L2learnerto mergethe phoneticpropertiesof LI and L2phonesthat havebeenidentifiedas belongingto the samein LI. Takenasa whole,theysuggestthatL2 learnerspro- category.Accordingto thishypothesis,learners'perceptualducenewL2 phoneslesswell than L2 phoneswith an L1targetfor L2 phonesmayevolveasa functionof experiencecounterpart.Howeverthe evidencewhichnow existsis inwith L2 (seeCaramazzaeta!., 1973;W'tiliams,1980),butsufficientto confirmor tfor L2 phonesmaynevermatchthosethe previousstudiesexaminedthe speechproductionofof L2 nativespeakersbecauseof interlingualidentification.talkerswithlittleor noknowledgeof theforeignlanguage Flege's(1984b}modelleadsto thepredictionthat,with suffifromwhichthe L2 phonesweretaken;mostexaminedmicientexperience,L2 learnerswill echproduction;andmostdifferentlyin L2 thanL1 (if the L1 andL2 stopsdifferphoreliedsolelyon phonetictranscription,oftenby non-native netically),but will neverperfectlymatchnativespeakersofspeaJ ers.L2. For example,if Englishlearnersjudge that the/t/ofWe applied perceptualand acousticcriteria to theEnglishand Frenchbelongto the samecategory,it is preFrenchsyllables/tu/("tous")and/ty/I"tu") to objectively dictedthey will persistin producingFrench/t/with relaexamineadults'productionof L2 phones.2English/u/ap- tivelylong (English-like)VOT valuesbecausetheir perceppearsto beproducedwith si nificantlylowerF2 valuesthantual target for French /t/ will merge the therhand,isaFrench/t/(including itsshort-lagVOT artin English.Con/t/(with its long-lagrOT values).trastiveanalysis(e.g.,Le Bras,1981)predictsthatL2 t L2erswillreplacea newL2 phonewitha "close"L1 phone,butlearnerswill be only partially successfulin producingL2offersno satisfactorymethodfor determiningwhichLIphones.Studieshaveshownthat learnerswhoseL1 realizesphoneisclosesttothenewL2phone.4Ifa newL2 phoneis/p,t,k/with short-lagrOT valuesproduceEnglish/p,t,k/consistentlyreplacedby a singleLI phonethe possibility with VOT valuesthat arelongerthanthosecharacteristicofexiststhat the L2 phonehas beenidentifiedwith that L1L1, butareneverthelesstooshortby Englishphoneticstanphone.Americanspeakersof FrenchoftenrealizeFrenchdards(Port and Mitleb, 1980;liege and Port, 1981;C.ara/y/as an/u/quality vowel(Oi udin,1953;Walz, 1979).Ifmazzaetal., 1973;Williams,1980;,cf.liege andHammond,boththe/y/and/u/of Frenchis identifiedwith English 1982).However,to our knowledgeno studyhasexamined/u/, nativeEnglishspeakersmight appearto produce theproductionof short-lagstopsin a targetforeignlanguageFrench/u/more accuratelythanFrench/y/since, in thatby learnerswhoseL1 realizes/p,t,k/with long-lagrOTcase,theywouldbe likelyto "substitute"English/u/forvalues.Kewley-Portand Preston(1974)hypothesizedthatboth vowels.Another possibilityis that native English short-lagstopsare lessdifficultto producethan long-lagspeakersdonotidentifyFrench/y/with anyvowelcategory stops.It is thereforepossiblethat native Englishspeakersof Englishandthereforelearnto produceFrench/y/withmay succeedbetter in producing the short-lag stops ofout referenceto speechpatternsestablishedfor the articulaFrench{C aramazzaand Yeni-Komshian,1974)than nativetion of English.If so, French/y/may be producedmoreFrench speakersproducethe long-lagstopsof English.IfaccuratelythanFrench/u/,at leastif adultsremaincapable AmericansaccuratelyproduceFrench/t/with theshort-lagof learningto producenewphones.VOT valuestypicalof Frenchmonolinguals,it woulddisIt seemslikelythattheamountof Frenchlanguageex- confirm the hypothesisthat ductionof French/u/and/y/bycreatesan absoluteupperlimit on the extentto whichL2learnersapproximatethephoneticnormsof a havefocusedonlanguage.speechtimingin L2 production.RelativelyexperiencedL2liege's {1984b}modelshouldapplyto vowelproduclearnershavebeenobservedto produceL2 tivelyinexper- tion as well as to the VOT dimensionin stop uracy709

Native Englishspeakersare likely to identifyFrench/u/with English/u/, despitethe fact that the lsareauditorilydetect-able(Flege,1984a).As a result,Englishlearnersof Frenchßmay developa perceptualtargetfor French/u/differingfrom that of Frenchmonolihgualsbecausethey mergethephoneticpropertiesof Frenchand English/u/. If so,theywill at bestproduceFrench/u/with formantvaluesintermediateto the uldthe productionof French/u/by Englishspeakersof Frenchbe observedto match that of Frenchmonolinguals,it wouldseriouslyunderminethe importanceof interlingualidentificationasa factorlimitingadults'successin producingL2 phones.Oneotheraspectof thepresentdatawill serveto testtheimportanceof egeandPort,1981) showthatthephoneticcharacteristicsof L1 phonesare oftenmaintainedin L2 speechproduction.To thebestof ourknowledge,nopreviousstudyhastestedthe effectof L2 learningon LI speechproduction.Ifidentifyingan L2 phonewith a phonein LI affectsthe per-ceptualtargetdevelopedfortheL2 phone,it shouldalsoaffectthe perceptualtargetpreviouslyestablishedfor the nersidentifyan L2 phonewith a phonein L1 theywill eventuallybeginproducingthe LI phonesothat it resemblesthecounterpartphonein L2. For example,a nativeFrenchspeakerwho identifiesEnglish/t/with the/t/of French shouldevent- lly beginproducingFrench/t/with VeT valuesthat exceedthe short-lagVeT valuestypicalof Frenchmonolinguals.A failureto noteanL2 effectonL ! speechproduction wouldalsotendto underminethe importanceof interlingualidentification.Our studyis dividedinto threeparts.In experiment1listenerslabelthe Frenchsyllables/tu/and/ty/producedbynative.French ndEnglishtalkers,If newL2phonesareproducedmoreaccuratelythan L2 phoneswhichhavea directcounterpartin L1, the/ty/produc l by ftenthan/tu/. Ifexperienceleadsto increasedL2 productionaccuracy,thesyllablesproducedby relativelyexperiencedEnglishspeakers of French shouldbe correctlylabeledmore often thansyllablesproducedby lessexperiencednativeEnglishspeakers. The effectof experienceis further exploredin experiment 2, which examinesthe identifiabilityof/ty/and/tu/in a iraired-comparisontask.Finally, we reportacousticmeasuresof VeT and formanes1-3 in the/tu/and/ty/syllables thatwereperceptually testedin experimentsI and 2. The predictedeffectof.interlingualidentificationis that French-Englishbilingualsand English-Frenchbilingualswill producethe/t/occurring in Frenchwordswith VeT valuesthat are intermediateto thoseobservedfor monolingualspeakersof EnglishandFrench.Further,nativeEnglishspeakersof Frencharepredictedto produceFrench/u/and with F 2 valuesintermedi-.ateto thoseobservedfor FrenchandEnglishmonolinguals.I. hichnative and non-nativetalkersproducedthe French syllables710J. Acoust.Soc. Am., VoL76, No. 3, September1984/tu/and/ty/. Accuracywasassessedby computingthefrequencywith .e.,asintended)thesesyllables.We examinedproduction of/tu/and/ty/inthree differentspeakingtaskstoensurea representativesamplingof Frenchspeechproduction.To examinetheeffectof linguisticexperience,we comparedtwo groupsof guageexperience.'A. Method 1. TalkersThree groupsof talkersdiffering'in languagebackground and experiencewere recruitedfrom a universitycommunityfor the with ggroup,designatedgroupNF, consistedof sixFrenchwomenandoneBelgianwomanwith ameanageof 38 years.Thesetalkershad livedan averageof12.2yearsin an dfourweremarriedto nativeEnglishspeakers.Thereweretwogroul sof nativeEnglishspeakerswhospokeFrenchasa secondlanguage,all womenfromtheMid-west.Thetalkersin eriegroup,designatedgroupNE- 1,consistedof undergraduatestudentswith a meanageof 22 yearswho had spentthe previousacademicyearin Paris.A moreexperiencedgroupof native Englishspeakers,designatedgroupNE-2, consistedof womenwitha meanageof 32yearswho held advanceddegreesin French and were teachingFrenchat an Americanuniversity.Talkersin thisgrouphadall spentseveralperiodsof timein France,the total averaging 1.3 years.One wasmarriedto a nativeFrenchspeaker.Talkersin the two Americangroups,like thenativeFrenchspeakers,wereusingEnglishas their primarylanguageatthe time of the study.Talkersin bothnativeEnglish-speakinggroupsbeganto studyFrenchin junior or seniorhighschoolbetweentheagesof 11and 17years.However,noneof themappearstohaveacquiredFrenchasa functionalsecondlanguagepriorto aboutage20. A languagebackgroundquestionnaireindicated that, comparedto talkers in group NE-1, thoseingroup NE-2 had substantiallymore formal instructioninFrenchlanguageand literature,rated their own productionand comprehensionof Frenchsomewhathigher,and usedFrenchsomewhatmoreoftenon a dailybasisin the periodimmediatelyprecedingthe experiment.No attemptwasmadeto objectivelyassessthe Frenchspeakingproficiencyof the two Americangroups,for theintentwassimplyto constitutetwoextremegroupsdifferingin overall experience.There was a clear differencebetweenthe two groupsin termsof the lengthof time theyhad usedFrench to communicate. For talkers in NE-Ithis was effec-tivelylessthana year,sincenoneof themhadusedFrenchon a regularbasissincetheir return from Paris 6 monthspreviously.The talkersin groupNE-2, on the other hand,had usedFrenchon a fairly regularbasisfor an averageofabout 10 years. Thus it seemsreasonableto refer to thetalkers in group NE-1 as "inexperienced,"and thoseingroup NE-2 as the relatively"expe enced"speakersofFrench.J.E. Flege and J. Hillenbrand:Limitson phoneticaccuracy710

2. SpeechmaterialsThefollowingtwosetsof phraseswereusedin counterbalancedorderto elicitproductionof the Frenchsyllables/tu/and/ty/by thenativespeakersof FrenchandEnglish:Tu !es al'canceto thefactthata singlesyllable(/tu/or/ty/} recurredat thebeginningof eachphrase.3. StimuliTousiespratresTousles(. vequesTu !esopposesmaterialfo perceptualanalysis(3groupsX 7talkersX 2syl-Tousles soldatsTu les observesTousles matinsTons les m&lecinsTu les renvoieslable typesX2 replicateproductionsX3speakingtasks).Phrases#4 mad#5 fromthetwophrasesetsweredigitizedat a 10-kHzsamplingratewith 12-bitamplituderesolution.The/tu/and/ty/syllables initiatingthesephraseswerethen isolatedusingthe segmentationcriteriaillustratedinFig. 1.The leftcursorwasplacedabout3 msto theleftof thesharpincreasein waveformenergysignalingthe releaseofconstrictionof/t/. The right cursorwasplacedat the zerocrossingthat wasjudgedto bestdemareatethe end of the"vowel" (/u/or/y/)and the beginningof the angesin waveformshapemadintensity,togetherwith a perceptualappraisal.Segmentationwasbasedon perceptualappraisalalonein the 5%10% of syllablesin whichno discontinuitywasvisuallyevidentin the at a edin this way was48.8 ms (s.d. 16)for groupNF, 44.8 ms (s.d. 15) for group NE-1, mad 51.1 ms(s.d. 20) for groupNE-2.Tousles dentistesTu !es obtiensTu I s informesTouslesgendarmesTu tu/or/ty/) inbothphrasesets vasheldconstant.The soundfollowingtheinitialsyllable(/1/) washeldconstantacrossthe two phrasesetsto makepossiblea directcomparisonbetween/tu/and/ty/. The final word in eachphrasetendedto receivecontrastivestressbecauseit variedacrossphrases.The twosetsof phraseswereproducedin threeprogressivelymore demandingspeakingtasks.The first task wassimplyto readthetestphrasesfoundin oneset,eachpreceded by the phrasenumberand a pause.The eachwithoneof thephrasesthathadjustbeenproducedin isolation.Productionwascuedby thesamewrittenlistof sevenphrasesusedin thephraseproductiontask.The finaltaskwasto producea storybasedonthe sevenphrases.The principalrequirementwasthat the storyineludea completesentenceinitiatedby eachof the sevenphrasesin theset.Thetalkersweregivena setof cards,eachbearingoneof thephrasesto beused.Theywereto arrangethe cards on the table before them in order to outline snecessaryto silentlyrehearse.They were permittedto intersperseadditionalsentencesnot initiated by one of the testphrasesin thestory,asneeded.The talkerswerenotrequiredto saythe numberof the phrasebeforeinitiatingeachuttermace,asin the previoustwo speakingtasks.However,theyweretold to pausebeforeeachsentenceto ensurethat thesyllableof interest(/tu/or/ty/} occurredin absoluteutterance-initialpositionin thestorytaskasit hadin the phraseA total of 252 syllableswereeditedfrom the speech4. Sub/ ctsThe subjectswere sevenfemale native speakersofFrenchwitha meanageof 38years.Sixhadservedastalkersabout3 mentindicatedthat noneof thesesubjectswereawarethat someof their own syllableshad beenincludedamongthe stimulipresented.5. ProceduresThe/tu/mad/ty/syllables werestoredona tionto listenersThe threespeakingtasksweremodeledusinga espeakingtasksusingtheseEnglishphrases.All butonetalker,whowasreplaced,wasableto performthethreespeakingtaskssatisfactorily.The storyproducedin thethird taskwashigh,ly naturaland spontaneousin the rerequiredtopausebeforeThe speech material was recorded (Sony modelTCDSM} in a sound-treatedroom with the experimenterseated about 5 ft from the talker. An eleotret condensor mi-crophone{N k michi modelCM-300) waspositionedabout6 in. from the talker'smouth.To counteractthe tendencyfortalkersto hyperarticulatein the presenceof a microphone{Labor, 1972),talkerswere told that the experimentexamined hat noneof the talkerswereawarethe experimentacttin11y711J. Acoust.Soc. Am., Vol. 76, No. 3, September1984J.E. Fiegoand d. Hillenbrand:Urnitson i: honetioaccuracy711

in a two-alternativeforced-choicetest. The syllableswereblockedon speakingtask(phrases,sentences,story)according to the phrase(: 4 or :5) in whichtheyhad beenprodueed. Within. a block, each syllable was presentedfivetimes.This yieldeda total of sixblocks,eachcontaining210stimuli (21 talkersX 2 syllablestypesX 5 ynormalizedfor overall rmsintensity.The task was to label eachstimulusas "tu" (/ty/) or"tous"(/tu/). The subjectswereinformedthat thesyllableshadbeeneditedfrom longerstretchesof speech,andthat anunspecifiednumberof thetalkerswerenotnativespeakersof/ty/syllables. Post-hoetests{Tukey'sHSD, alpha-----0.01)revealedthat the /tu/ syllablesproducedby the nativeFrenchtalkers{NF} werecorrectlyidentifiedmorefrequently than the /tu/ syllablesproducedby the experiencedAmericantalkers(NE-2) who,in turn, produced/tu/syllablesthat werecorrectlyidentifiedmoreoftenthan thoseoftheinexperiencedAmericans(NE-I). Post-hoctestsrevealedthat for/ty/, on the otherhand,syllablesproducedby thenativeFrenchtalkers(NF) were correctlyidentifiedmorefrequentlythan syllablesproducedby talkersin the twoAmericanspeakergroups(NF,-I, NE-2), but that therewasno differencebetweenthe two Americangroups.French.The meanpercentcorrectidentificationscoresfor syllaThe percentageof timeseachsubjectcorrectlyidentiblesproducedby individualtalkersin the three speakerfiedthe/ty/and/rod din TableI. In thistabledatahavebeeneachof thethreespeakergroups(NF, NF,-1,NF.-2)in tory).The maxispeakingconditions.The data for individualtalkersweremum numberof correctidentificationsof/ty/and/tu/wasanalyzedin a mixeddesignanalysisof variancein which70 (7 talkersX 2 replicateproductionsX 5 presentations).Inspeakergroupwas a between-groupfactor, and speakiqgaddition,thepercentageof times/tu/and/ty/syllables protaskand syllabletypewererepeatedmeasures.ducedby eachof the21 talkerswerecorrectlyidentifiedwasIn this analysisthe effectof speakergroupwas againcomputedseparatelyfor eachspeakingtask.The maximum significant[F(2,18)--20.2, p 0.001]. Post-hoctests renumberof correctidentificationswas again 70 (7 listen- yealedthat the nativeFrenchtalkers(NF) producedsyllaersX 2 replicateproductionsX 5 presentations).blesthat were identifiedmore correctlythan thoseof theexperiencedAmericans(NE-2)who,in turn,producedmoreB. Resultsidentifiablesyllablesthan the inexperiencedAmericans(NE4) (p 0.01).Syllablesproducedby the nativespeakersof FrenchThe interactionbetweenspeakergroup syllabletype{NF) werecorrectlyidentifiedmoreoften{about95% of thewas not signticantas it was in the "listener" analysistime) than syllablesproducedby either the experienced[F(2,18) 2.11,p 0.145].Only sixof sevennativeFrenchAmericans(about75% correct)or the inexperiencedAmeritalkers,and fiveof sevenexperiencedAmericansproducedacans(about60%}. The/tu/syllables producedby the nativemore identifiable/tu/than/ty/. Only six of the seveninexFrenchand experiencedAmericanswere correctlyidentiperiencedAmericansshowedthe oppositepattern,producfied at a somewhathigher rate than/ty/. For the inexperingamoreidentifiable/ty/than/tu/.ienced American talkers, on the other hand,/tu/was corThe effectof speakingtaskdid not reachsignificancerectly identifiedat a substantiallylower rate {about45%[F(2,36) 2.43],but the ut 70% correct).As a result,therewasandsyllabletypedid[F(2,36) 9.04,p 0.001].Thesyllablelittle differencebetweenthe experiencedand inexperienced/tu/tended to becorrectlyidentifiedmoreoftenthan/ty/inAmerican talkers for /ty/, but a substantialdifferencethe phraseand sentencetasks,whereasthe reversewastruebetweenthesetwo groupsfor/tu/.in thestorytask.However,testsof simplemaineffectsindiThe percentcorrectidentificationscoreswere transformedusingan arcsinetransformation(Kirk, 1968,p. 66)becausehomogeneityof variancecannotbe assumedwhen TABLE I. The percentageof times/tiff and/ty/syllables producedbyan analysisisbasedon percentages.The transformedscores nativespeakersof French(NF}, experiencedAmericanspeakersof French{NE-2),andinexperiencedAmericanspeakersof French(NF I) werecorweresubmittedto a three-wayanalysisof variancein whichrectlyidentified.Eachsc,a e isbasedona totalof 210forced-choiceiden,tifispeakergroup{NF, NF ,-I,NF ,-2),speakingtask (phrases, cafions(7 listen( 'sX 2 replicateproductionsX 3 speakingtasksX 5 presensentences,story),and syllabletype (/tu/or/ty/)were alltations).repeatedmeasures.SpeakergroupThe effect of speaking task was not significantNFNE-2NE-I(p 0.01).However,the interactionof speakergroupX hlysignificant[F(2,12) 51.27,p 0.001].I99.595.192.382.840.361.3Testsof simplemaineffectsindicatedthat the/tu/syllables299.591.589.066.189.53.8producedby talkersin the nativeFrench (NF) and exper398.597.535.884.141.294.1iencedAmerican (NE-2) groupswere correctly .123.669.0more often than/ty/syllables, oducedby the inexperiencedAmericans(group NE-1)798.092.684.397.050.287.1were correctly identifiedless often than /ty/ syllables(p 1.9)(25.1)(22.1)(32.0)Testsof simplemain effectsalsoindicatedthat the effect of speakergroupwassignificantfor both the/tu/and712J. Acoust.Soc. Am. VoL70. No. 3. September1984J.E. Fiegoand J. Hillonbrand:L'mitson phoneticaccuracy712

primarilybycated that this interactionwas of marginalimportance. thatFrench/y/and/i/are finesintermsoftheTherewasno significanteffectof syllabletype(?tu/versuspositionof thetonguedorsum./ty/) in anyof thethreespeakingtasks,andtheeffectof anterior-posteriorSupportingthisaredatapresentedby rthe/tu/or protrn/ty/syllables (p 0.01).sionbetweenFrench/i/and/y/, littledifferencein theareaC. Discussionof the mouthorifice,andlittle differencein theratioof theto theverticalopeningof themouthduringproIt is not surprisingthat listenerscorrectlyidentified enreportedmoreFrenchsyllablesproducedby FrenchthanAmericanfor Dutch vowels.Basedon acousticanalysis,EMG data,talkers.Whatissomewhatsurprisingistheextentto whichbyarticulatorysynthesis,Raphaeletal. experienced ,althoughtherewasnodifferencein tyto producea syllablecontaininga "new"vowel(i.e.,/ty/} doesnotbenefitfromadditionalL2 experience,whereastheabilityto producea syllablewitha vowelthathasa counterpartin L1 (i.e.,/tu/) doesshowaneffectofadditionalexperience.7An examinationof syllablesproducedby production.Theinteractionof speakergroupX syllabletypedid not reachsignificancein the"talker"analysisbecause4 of 21 talkersfailedto conformto thegeneralpatternevidentin n thecorrectidentificationof/tu/and/ty/because /ty/, whereastheinexperiencedtalkerstendedto produce/ty/betterthan/tu/. Thusthedatadonotsupportthehypothesisthata newL2 vowelwhichhasno directcounterpartin L1 (e.g.,/y/}willbeproducedmoresuccessfullythana vowelwhichdoeshavesucha counterpart(e.g.,/u/).Onlytwoof thesevenexperiencedAmericans,andtwoof sthatwerecorrectlyidentified93% atchnativeFrenchspeakersin theabilityto produceFrench/y/.Theirfailureto dosomaybetheresultof in Americanschoolsareexplicitlytaughtto produceFrench/y/by han/y/.Theuseof ch/y/ mightalsoresultin ccordingtoLinker{1982),French/y/ is producedwith nEnglish/u/, andhasa substan

phones (speech sounds) found in a foreign language. Researchers {e.g., Lado, 1957) have frequently observed that L2 learners tend to produce the words of a foreign lan- guage with phones that can be heard in the phonetic surface ofLl. Weinreieh {1953) hypothesized that this resulted from the "interlingual identification" of L1 and L2 phones. Vald-

Related Documents:

Phonetic Transcription Articulation of Sounds Phonetic Alphabet Transcription Transcription Notes Phonetic transcriptions are written in square brackets [ ]. Transcribe words based on sound, not spelling. Don’t use a schwa ([@]) in stressed syllables. Upper- and lowercase letters are not interchangeable. Some morphemes, like past tense -ed .File Size: 2MB

Unit II. Limits (11 days) [SC1] Lab: Computing limits graphically and numerically Informal concept of limit Language of limits, including notation and one-sided limits Calculating limits using algebra [SC7] Properties of limits Limits at infinity and asymptotes Estimating limits num

Aug 26, 2010 · Limits of functions o An intuitive understanding of the limiting process. Language of limits, including notation and one-sided limits. Calculating limits using algebra. Properties of limits. Estimating limits from graphs or tables of data. Estimating limits numerically and

Learn the International Phonetic Alphabet (IPA) Learn the Sounds of English with the IPA – Sample Lesson Plan For more fun worksheets, games and quizzes log onto www.englishbanana.com now! Talk a Lot Elementary English Banana.com 18.4 “If you know the phonetic alphabet, you know how to pronounce words. This alphabet gives

Phonetic Diversity, Statistical Learning, and Acquisition of Phonology* Janet B. Pierrehumbert Northwestern University 1Introduction Infants show evidence of phonetic categorization and of perceptual parsing of the speech stream before they learn to speak, before they have large vocabularies, and

natural phonetic variability within a phonetic . diversity in language learning, it is imperative to increase the range of languages. Extending previous research, therefore, the current study examines how . Statistical analysis The perception performance of Korean codas was

McGurk fusions across phonetic contexts ABSTRACT The McGurk effect has generally been studied within a limited range of phonetic contexts. With the goal of characterizing the McGurk effect through a wider range of contexts, a parametric investigatio

1 Phonetics: A “Sound” Science 1 Learning Objectives 1 Phonetics and the International Phonetic Alphabet 1 Variation in Phonetic Practice 3 The IPA and Unicode Fonts 4 Chapter Summary 6 Study Questions 6 Online Resources 7 2 Phonetic Transcription of English 9 Learning O