Variation in contrastive phonation in Santa Ana Del Va - PDF document

ChristinaM.Espositolanguages/dialectssuchasJapanese(Todaka1993),AmericanEnglish(Klatt&Klatt1990,Hanson1997,Hanson&Chuang1999),andBritishEnglish(Henton&Bladon1985).Someofthesegender-dependentvariationscouldbeduetophysiologicaldifferences;forexample,directobservationsofthevocalfoldsviaÞberscopyshowedthatfemalesweremorelikelytohaveincompleteglottalclosurethanmales,whichcanproduceabreathiervoicequalityodersten&Lindestad1990).Gender-basedphonationdifferences,however,donotalwaysvaryinasystematicway.Forexample,whileresultsofacousticanalysesshowedthatonaverageAmerican-EnglishÐspeakingfemaleswerebreathierthanmales,therewasagreatdealofgendervariationand,infact,somemaleswerebreathierthanthefemales(Klatt&Klatt1990).Furthermore,Redi&Shattuck-Hufnagel(2001)examinedglottalization(aphonationsimilartocreaky)andfoundthatglottalizationinAmericanEnglishcouldnotbepredictedfromgenderalone.Inadditiontogender,studieshavesuggestedthatphonationcanbesensitivetochangesinfundamentalfrequency(F0),whetheratthelexicalorutterancelevel.Forexample,alowerF0wascorrelatedwithbreathy-voicedstopsinHindi(Ohala1973),whileahigherF0wascorrelatedwithatensevoiceinJingpho,LahuandYi(Maddieson&Hess1987).Furthermore,Hagen(1997)foundthatinEnglishandGerman,glottalizationoccurredmoreoftenonwordsproducedwithalowF0.Studieshavealsoshownthatphonationcanvaryduetoposition.Forexample,Epstein(2002)showedthroughinverseÞlteringthattensephonationswerefoundutterance-initiallyregardlessofF0inAmericanEnglish.OtherstudieshaveshownthattheendsofsentencesorparagraphsinEnglishcanbeassociatedwitheither(i)creakiness(orglottalization)(Lehiste1975,Kreiman1982,Henton&Bladon1985)or(ii)aÔbreathy-laryngealizedÕvoice,atypeofvoicequalitycharacterizedbyasimultaneousincreaseintheposteriorglottalchinkcreatingbreathyvoiceandarotationoftheanteriortipsofthearytenoidcartilagescreatinglaryngealization(Klatt&Klatt1990).(Thiseffectcanlikelybeexplainedastheinitiationofaryepiglottictrilling,Esling2005.)Severalstudieshavereportedcreakphrase-initially,phrase-Þnallyandattheboundaryofsmallerprosodicunitsfromdirectmeasuring/observationofthevoicingsource(Pierrehumbert&Talkin1992,Pierrehumbert1995)orthroughqualitativeassessmentandlabelingofwaveforms(Dilley,Shattuck-Hufnagel&Ostendorf1996,Hagen1997).Additionally,theendsofsentenceswereassociatedwithalaxvoicequalityinSwedish(Gobl1988,usinginverseÞltering),andacreakyvoicequalityinFinnish(Ogden2003,obtainedthroughqualitativeassessmentandlabelingofwaveforms).WiththeexceptionofEpstein(2002),however,itisdifÞculttotellfromthesestudiesifitisphonologicalpitchorpositionthatalteredEachoftheaforementionedstudies(withtheexceptionofHindi,Ohala1973)examinedvariationinphonationinalanguagewitheitherallophonicorsuprasegmentalnon-modalphonation.Lessattentionhasbeengiventothissortofvariationinlanguageswithphonemicphonationcontrasts,thoughsomegender-dependentdifferenceshavebeenobserved.(Nostudyhasexaminedorobservedchangesinphonationduetosentence-levelF0and/orpositioninlanguageswithcontrastivephonation.)InJalapaMazatec,alanguagethatcontrastsbreathy,modalandcreakyvowels,femalesweresigniÞcantlybreathierthanmalesintheiracousticmanifestationofbreathiness(Blankenship1997)andinSanLucasQuiavinõZapotec,alanguagethatcontrastsbreathy,modal,creaky,andcheckedvowels(avowelfollowedbyaglottalstopwithaphonationtypedistinctfrommodalandcreaky,Munro&Lopez1999),Gordon&Ladefoged(2001)observedspectrographicevidencethatthefemalespeakerwasbreathierthanthemalespeaker.Whilethereissomeevidencethatphonationwillvary,atleastasafunctionofgender,littleisknownaboutpotentialsourcesofvariationinlanguageswithcontrastivephonation.Thepresentstudysetsouttoinvestigatevariationinphonationdueto(i)gender,(ii)F0,and/or(iii)prosodicpositioninSantaAnadelValleZapotec,anOto-Mangueanlanguagewithphonemicallybreathy,modal,andcreakyvowels. ContrastivephonationinSantaAnaDelValleZapotecBeforediscussingthepresentstudyindetail,therewillÞrstbeapresentationofbackgroundinformationonSantaAnadelValleZapotec,thenareviewoftherelevantacousticmeasuresofphonation,followedbymethodology,resultsanddiscussionoftheBackground2.1AboutthelanguageSantaAnadelValleZapotec(hereafterSADVZ)isanOto-MangueanlanguagespokeninSantaAnadelValle,Oaxaca,Mexico.TheEthnologue(R.Gordon2005)classiÞesSADVZasbelongingtotheSanJuanGuelavõaZapotecsubgroup,whichcontainsthenumerousanddiverselanguagesspokenintheValleyofOaxaca,suchasSanJuanGuelavõa(forwhichthesubgroupisnamed),JaliezaZapotec,TeotitlandelValleZapotec,andSanMartõnTilcajeteZapotec,tonameafew.Thereareapproximately28,000speakers(1990census)fortheentireSanJuanGuelavõaZapotecsubgroup;itisnotknownwhatportionofthisiscomposedofSADVZspeakers.SADVZhassixvowels[ ],eachofwhichcanhaveoneofthreecontrastivephonations:modal,breathy,orcreaky.Inaddition,toneiscontrastiveonmodalvowels,whichcanhaveeitherahighorarisingtone.Thereisnodifferenceinphonationbetweenthehighandrisingmodalvowels.Breathyandcreakyvowelsbothhaveafallingtone(Esposito2003,2004a).Thus,thereisastrongrelationshipbetweenlexicalF0andphonationinthatmodalphonationsareonlyproducedwithhighF0s.Thequestionofwhethersentence-levelF0willhavethesamerelationshipwithphonationwillbeexploredinthecurrentstudy.SantaAnadelValleZapotec,isa(V)erbÐ(S)ubjectÐ(O)bjectlanguage.AnexampleofabasicdeclarativesentencewithVSOwordorderisgivenin(1).(1)g-seeElenafrogÔElenasawafrog.ÕHowever,SADVZalsoallowsSVOandOVSwordorder(inwhichcasethepreverbalmaterialhasafocusedreading).TherelativelyfreewordorderofSADVZmakesitanideallanguageinwhichtoexaminetheeffectsofpositiononphonationbecausewordscanbeelicitedinavarietyofpositions.Furthermore,thereisnoutterance-Þnaldriftinsentence-levelF0,makingitpossibletostudytheeffectsofpositionwithoutastronginßuenceofsentence-levelF0.AstudyoftheintonationofSADVZshowedthatinsentence-Þnalpositioninadeclarativesentencetheoriginaltonalpatternwaspreservedthoughtheoverallpitchvaluewaslowered.Forexample,ifalexicallyhightonewordwasinsentence-Þnalpositioninabasicdeclarative,therewasaphonologicalloweringoflexicalF0,butthelevelcontourwaspreserved(i.e.itdoesnotchangetoafallingF0).ThisisexempliÞedinFigure1,apitchtrackofthesentenceinexample(1);theendofthesentenceendswithalow-levelF0,andnotafallingF0.InFigure1andthroughout,theSADVZwordsarewritteninIPA(withtheexceptionofthewhichrepresents[])onthewordtier,followedbythelexicaltonerisingtone,Hhightone),glossandtranslation.Itisalsopossibletoexaminethepotentialinßuenceofsentence-levelF0onphonation,independentlyofposition,throughtheexaminationofavarietyofintonationalcontours.Forexample,wordsinsentence-medialpositionhavethesamesentence-levelF0aswordsutteredinisolationwithoutfocus.NotethesimilarityintheF0ofÔElenaÕ[]aspronouncedin ThepitchtrackspresentedinthispaperarefromEsposito(2004b). ChristinaM.Esposito Figure1Pitchtrackof[]‘Elenasawafrog’. Figure2Pitchtrackof[]‘Elena’producedinisolationwithoutfocus.isolation(withoutfocus)(Figure2)tothewayinwhichitispronouncedinsentence-medialposition(Figure1).Theindependentrelationshipbetweensentence-levelF0andposition,therelativelyfreewordorder,andthelackofanutterance-ÞnaldriftinF0makeSADVZanideallanguageinwhichtostudyvariationinphonation.2.2AboutmeasurementsTherearenumerousacousticpropertiesthatcanbeusefulmeasuresofphonation,thoughspectralmeasureshavebeenthemostpopularmethodformeasuringphonationfromanaudio ContrastivephonationinSantaAnaDelValleZapotecsignal.SpectralmeasureshavebeenareliablemeasureofphonationinHmong(Huffman1987,Andruski&Ratliff2000),Mazatec(Kirk,Ladefoged&Ladefoged1993,Silvermanetal.1995,Blankenship1997),and!Xo(Bickley1982,Ladefoged1983,Ladefoged,Maddieson&Jackson1988),tonameafew.Primarily,thedifferencebetweentheamplitudesoftheÞrstandsecondharmonics(H1-H2)hasbeenusedtodistinguishphonation(e.g.Fischer-J¿rgensen1967;Bickley1982).Otherstudies,however,havemadeuseoftherelationshipbetweentheamplitudeofH1comparedtotheamplitudeofharmonicsexcitinghigherformants(e.g.A1,A2,A3,andA4).Theseinclude:H1-A3(Stevens&Hanson1995,Blankenship1997),H1-A1orH1-A2(Ladefoged1983)andtheaverageofH1-H2comparedtoA1(Stevens1988).Themajorityofstudiesonlinguistically-relevantvoicequalities(e.g.Bickley1982,Ladefoged1983,Huffman1987,Ladefogedetal.1988,Kirketal.1993,Silvermanetal.,1995,Blankenship1997)havenotmadeuseofcorrectedornormalizedmeasures,butinsteadfocusedon//,becausethehighÞrstformantminimizestheeffectsontheÞrstandsecondharmonics.(Andruski&Ratliff2000alsouseduncorrectedmeasures,butexpandedtheirstudyto//and//inadditionto/Thevariousspectralmeasureshavebeenassociatedwithphysiologicalcharacteristics.Holmbergetal.(1995)showedthatthedifferencebetweentheÞrstharmonic(H1)andthesecondharmonic(H2)correlatedwiththeproportionoftheglottalcycleduringwhichtheglottisisopen(theopenquotient).Whenthevibrationofthevocalfoldshasalargeopenquotient,theamplitudeofH1isgreaterthantheamplitudeofH2.Forexample,breathyphonationhasalargeopenquotientandaspectrumdominatedbyH1.Furthermore,Stevens(1977)suggestedthatmeasuresofspectralslopecorrelatedwiththeabruptnessofvocalfoldclosure.Onewaythatvocalfoldvibrationcanbeachievediswithtightlyadductedarytenoids,allowingvibrationovertheanteriorportionofthevocalfolds.AsStevens(1977:274)notes,Ô[w]ithinthisregion,themechanicalpropertiesofthefoldsaremoreuniform,andanabruptclosurealongthelengthofthevibratingportioncanbeexpected.Amorerapidrateofclosureisalsoexpected,sincetheinward(adducting)forceonthefoldsisgreaterwhenthearytenoidsaretightlyadductedÕ.ThisconÞgurationproduceswaveformswithmorehigh-frequencyenergy.Forthisreason,increakyphonation,whichcanbecharacterizedbyvocalfoldsthatcloserapidly(butalsoopenmoreslowly),theamplitudeofthehigherharmonicsofthevowelisgreaterthanthatofthefundamental.VariationinphonationGendervariationwillbeexaminedÞrst,followedbyF0andposition.3.1GendervariationTodetermineifthereisapossiblegenderdifferenceinphonationinSADVZ,malesÕandfemalesÕphonationswerecomparedandmeasuredusingtwotypesofspectralmeasures,onereßectingtheopenquotient(H1-H2),theotherreßectingthespeedofvocalfoldclosure3.1.1Methods3.1.1.1SpeakersFivenativespeakersofSADVZ(threemaleandtwofemale)wereselectedforthisstudy.Speakersrangedfrom40to70yearsofage.Speaker1isatrilingualmale,speakingSADVZandSpanishnatively,inadditiontoEnglish,whichhelearnedinhislatetwenties.Speakers2and3arebilingualmalesspeakingSADVZandSpanishnatively.Speaker4isabilingual Apilotstudy(Esposito2003)comparedsixmeasuresanddeterminedthatH1-A3wasthebestspectraltiltmeasureinSADVZ. ChristinaM.EspositoTable1SantaAnadelValleZapotec(SADVZ)wordlist.Thefortis/lenisobstruentsarerepresentedwiththesymbolsforvoicelessandvoicedconsonants,respectively.ThisistheirtypicalrepresentationinZapoteclanguages.Thefortis/lenissonorantsarerepresentedbytheirlengthcontrast. SADVZGLOSSSADVZGLOSSSADVZGLOSS pa!p]‘potato’[]‘four’[]‘hidden’hidden’pa!g]‘pay’[]‘Tlacolula’[]‘tostick’’fa!b]‘detergent’[]‘price’[]‘eld’’ta!p]‘lid’[]‘tepache’[]‘girl’’la!t]‘(tin)can’[]‘care’[]‘rough’rough’da#d]‘father’[]‘chocolate’[]‘scabies’scabies’ba#d]‘duck’[]‘clothes’[]‘butt’butt’fa#ld]‘skirt’[]‘thick’[]‘skunk’skunk’ma#kw]‘Mark’[]‘cold(n.)’[]‘have’ve’la#s]‘skinny’[]‘mountain’[]‘empty’ femalespeakingSADVZandSpanishnatively.Speaker5isamonolingualSADVZ-speaking3.1.1.2SpeechmaterialsSpeakerswereaskedtoproduce10monosyllabic[a]-vowelwordsperphonationtype(breathy,modal,andcreaky)foratotalof30tokensperspeaker(seeTable1forthewordlist).(Formodalphonation,thisincludedÞvetokenswithahightoneandÞvewitharisingtone.Previousresearchshowedthattherewasnotaphonationdifferencebetweenthehighandhigh-risingtones;seeEsposito2003.)Onlywordswith[a]wereselectedbecausetheÞrstformantoflowvowelsdoesnotinßuencetheamplitudeoftheÞrstorsecondharmonicsasmuchasinhighervowels.Eachtokenwasutteredinsentence-medialpositionintheframe[g]ÔSay___Þrst.Õ,andrepeatedtentimesbyeachspeaker.Itwasnotpossibletocontroltheonsetandcodaconsonantsofthetargetwords.However,pilotresearchsuggestedthatthecodaconsonantsdidnothaveaneffectonphonation(onlycodaconsonantsweretestedbecauseonlytheendofthevowelswasmeasured).Pairedt-testsindicatedthatthephonationofvowelsproducedbeforelenisconsonantswasnotsigniÞcantlydifferentfromthephonationbeforefortisconsonants(df58,t.14,p3.1.1.3ProcedureSpeakerswererecordedovermultiplesessionsinasoundproofboothattheUCLAPhoneticsLab.TokensweredigitizedandanalyzedinPCQuirer(SciconR&D,Inc.,Encino,CA)atasamplingrateof22,050Hz.Eachvowelwasdividedintofourequalparts.Twoacousticmeasures,H1-H2andH1-A3,weremadeforeachvowelovera30mswindowinthelastquarterofthevowel(approximately50ms).PreviousresearchshowedthatthephonationcontrastswerelocalizedtotheendofthevowelinSADVZ(Esposito2003).Spectrogramswereusedtopositionthe30mswindow.MeasurementsweretakenfromaFastFourierTransform(FFT).Atotalof1500tokenswereanalyzed(5speakers30wordsrepetitionseach3.1.2ResultsFigures3,4,5,6,and7aregraphsoftheaverageH1-H2andH1-A3valuesperphonationtype(breathy,modal,andcreaky)forSpeakers1,2,3,4,and5,respectively.Thedifferencebetweentheamplitudesoftheharmonicsisgivenonthey-axisindB.GreaterdBvaluesindicatemorebreathiness;smallerdBvaluesindicatemorecreakiness.IntheÞguresinthissectionandthroughout,anarrowispointinginthedirectionofincreasedbreathinessandaveragesaregiveneitherabove(forpositivevalues)orbelow(fornegativevalues)eachcolumn.BeloweachÞgure,therangeofvaluesandstandarddeviation(S.D.)ispresented. ContrastivephonationinSantaAnaDelValleZapotec Figure3AverageH1-H2andH1-A3values(indB)forSpeaker1(male).Therangeandstandarddeviationsarepresentedbelowthegraph. Figure4AverageH1-H2andH1-A3values(indB)forSpeaker2(male).Therangeandstandarddeviationsarepresentedbelowthegraph.Formalespeakers,H1-A3distinguishedthethreephonemicphonationsintheexpecteddirections(i.e.withbreathyphonationhavingthehighestdBvalue,followedbymodalandthencreaky),butH1-H2didnot.Forallthreemalespeakers,theH1-A3valuesforcreakyphonationwereconsistentlynegative.Thecreakyvowelsforthemalespeakerswereremeasuredusing(whereindicatesthatthemeasurewascorrected)usingVoiceSauce(Shue,Keating&Vicenik2009).ResultswerenotsigniÞcantlydifferent(p.05)andwerestillnegative. PreviousstudiesusingH1-A3(orH1,wheretheasteriskindicatesthatthemeasurewascorrected)havenotreportednegativevaluesforthismeasure.Forexample,Hanson&Chuang(1999)reportedonlypositivevaluesforH1formaleEnglishspeakersproducing[],[¾]and[].For[],thevaluesrangedfrom5.7dBto23.1dB;for[¾],6.2Ð24.1dB;andfor[],4.8Ð22.8dB.Inaddition,Stevens&Hanson(1995)alsoreportedpositivevaluesforH1forfemaleEnglishspeakersproducing[],[¾]and[].For[],thevaluesrangedfrom17.9dBto35.2dB;for[¾],18.0Ð39.1dB;andfor[],14.1Ð33.5dB.WhilepreviousstudieshavenotreportednegativevaluesforH1-A3(orH1),Blankenship(1997)reportednegativevaluesforasimilarmeasure,H1-A2,forlaryngealizedvowelsinMazatec. ChristinaM.Esposito Figure5AverageH1-H2andH1-A3values(indB)forSpeaker3(male).Therangeandstandarddeviationsarepresentedbelowthegraph. Figure6AverageH1-H2andH1-A3values(indB)forSpeaker4(female).Therangeandstandarddeviationsarepresentedbelowthegraph.Figure8presentsawaveform,spectrogram,andFFTfor[]ÔÞeldÕproducedbySpeaker2.NotehowtheamplitudeofA3ishigherthanH1.Forfemalespeakers,H1-H2successfullydistinguishedthethreephonemicphonationcategoriesintheexpecteddirections,whileH1-A3didnot.Unfortunately,itisdifÞculttodetermineifmaleSADVZspeakersarebreathier/creakierthanfemales(andviceversa)becauseofthesuccessofthedifferentmeasures.However,thesuccessofH1-A3forthemalespeakersandH1-H2forthefemalespeakerssuggeststhatmenandwomenareusingdifferentlaryngealsettingstoproducethesamelinguisticcontrast,thoughdirectobservationsofthevocalfoldsareneededtosupportthisclaim. ContrastivephonationinSantaAnaDelValleZapotec Figure7AverageH1-H2andH1-A3values(indB)forSpeaker5(female).Therangeandstandarddeviationsarepresentedbelowthegraph. Figure8Waveform,spectrogramandFFT(calculatedovera30-mswindow)of[]‘eld’producedbySpeaker3.Pitchtrackfailsattheend.3.2ProsodicvariabilityTodetermineifSADVZphonationissensitivetochangesinsentence-levelF0and/orposition,acousticmeasures(H1-A3forthemalespeakersandH1-H2forthefemalespeakers)weretakenofthesamewordsinÞveprosodicpositions,eachassociatedwithanF0.Theeffectsofsentence-levelF0andpositionwillbeteasedapartintheresultssection. ChristinaM.Esposito3.2.1Methods3.2.1.1ProcedureThesamespeakersasthoseusedintheprevioustaskwereaskedtoproduceten[a]-voweltokensperphonation(aswiththepreviousexperiment,thisincludedÞvetokenswithahightoneandÞvewitharisingtoneformodalphonation)inthefollowingÞveprosodic(i)isolation(withfocus),whichhasahigherF0thansentence-medialposition(ii)initialposition(focused),whichhasahigherF0thansentence-medialposition(iii)isolation(withoutfocus),whichhasamid-rangeF0(iv)medialposition,whichhasamid-rangeF0(v)Þnalposition(endofadeclarativesentence),whichhasalowerF0thansentence-medialIsolation(withfocus)waselicitedbyaskingthespeakersaquestiontowhichtheresponsewasthetargetword.Isolation(withoutfocus)waselicitedbyaskingthespeakerstosimplysaythewordinisolation.Wordsutteredinisolationandnotasananswertoaquestionwerealwaysproducedwithoutfocus.H1-A3andH1-H2weremeasuredfromanFFTfollowingthesameprocedureestablishedabove.Inaddition,F0wasmeasuredatthreetime-points(beginning,middle,andend).3.2.2ResultsFigures9,11,and13aregraphsoftheaverageH1-A3valueforeachphonationcategoryinÞveprosodicpositions(isolationwithfocus,initial,isolation(notfocused),medial,andÞnalposition)forSpeakers1,2,and3,respectively.Figures15and17aregraphsoftheaverageH1-H2foreachphonationcategoryinÞveprosodicpositions(isolationwithfocus,initial,isolation(notfocused),medial,andÞnalposition)forSpeakers4and5,respectively.Thedifferencebetweentheamplitudesoftheharmonicsisgivenonthey-axisindB.Therangeandstandarddeviationsarepresentedunderthegraphs.Figures10,12,14,16,and18aregraphsoftheaverageF0(inHz)inÞveprosodicpositions(isolationwithfocus,initial,isolation(notfocused),medial,andÞnalposition)forSpeakers1,2,3,4,and5,respectively.InthegraphsoftheF0,thehighandrisingtonemodalvowelsareInallÞveprosodicpositions,thereisathree-waycontrastinphonation.Thiscontrast,however,isnotalwayswell-deÞned.Inisolation(mid-rangeF0),medialposition(mid-rangeF0)andÞnalposition(lowerF0),thethree-wayphonationcontrastisclearestforboththemaleandthefemalespeakers(butwithsomechangesforthefemalespeakers,whosemodalvowelswerecreakierword-Þnallythaninmedialposition).Thecontrastwasminimizedinisolationwithfocus(highF0)andinitialposition(highF0).Inthesetwopositions,breathyandcreakyvowelshadamuchmoremodalphonationthanwhenproducedinisolation(mid-rangeF0),medialposition(mid-rangeF0)orÞnalposition(lowerF0).ThereisevidencethatitisF0,independentofposition,thatisinßuencingphonationinSADVZ.Tokenswiththesameposition,butdifferentF0s,havedifferentphonations.MorespeciÞcally,tokensinisolationcanbeproducedwitheitherahighF0(whenfocused)oramid-rangeF0(whennotfocused).ItwasonlywhenthetokenswereproducedwiththehighF0,however,thatthethree-waycontrastinphonationwasminimized.Furthermore,tokens Wordsinisolationcanbeelicitedwithtwosentence-levelF0s;onehighandonemid-range.Whiletechnicallynottwoseparatepositions,bothwereelicitedbecausetheywerehelpfulindeterminingifF0inßuencesphonationindependentlyofposition. ContrastivephonationinSantaAnaDelValleZapotec Focus (high F0) RangeS.D.RangeS.D.RangeS.D.RangeS.D.RangeS.D. 1 to 7 2.01 1 to 7 2.10 11 to 18 3.19 11 to 17 2.80 18 to 23 1.87 -1 to 5 2.90 -1 to 5 2.03 1 to 6 2.13 -1 to 7 2.98 3 to 8 2.31 -3 to -8 2.95 -2 to -5 2.50 -4 to 1 2.26 -5 to 0 2.87 -11 to Ð Figure9AverageH1-A3valuesforbreathy,modal,andcreakyphonationinveprosodicpositionsforSpeaker1(male).Therangeandstandarddeviationsarepresentedbelowthegraph. Figure10AverageF0forSpeaker1(male)inveprosodicpositions(isolationwithfocus,initial,isolation(notfocused),medial,andnalposition)atthreetimepoints(1,2,3)pervowel.withthesameF0butdifferentpositionshavesimilarphonations:inpositionswithmid-rangeF0(i.e.isolation(non-focused)andsentence-medialposition),therewasaclearthree-waycontrastinphonation. ChristinaM.Esposito Focus (high F0) (mid F0) Range S.D. Range S.D. Range S.D. Range S.D. Range S.D. -1 to 8 2.37 -1 to 5 2.15 8 to 13 1.91 9 to 14 2.00 15 to 18 2.43 -2 to 3 2.15 -1 to 2 1.97 1 to 5 2.31 1 to 5 2.21 3 to 7 2.15 -1 to 2 1.97 -2 to 1 1.87 -6 to 0 2.11 -5 to 0 2.01 -10 to 0 2.81 Figure11AverageH1-A3valuesforbreathy,modal,andcreakyphonationinveprosodicpositionsforSpeaker2(male).Therangeandstandarddeviationsarepresentedbelowthegraph. Figure12AverageF0forSpeaker2(male)inveprosodicpositions(isolationwithfocus,initial,isolation(notfocused),medial,andnalposition)atthreetimepoints(1,2,3)pervowel.DiscussionandconclusionThecurrentstudysetouttodetermineifphonationwouldvaryasafunctionof(i)gender,(ii)F0,and/or(iii)positioninSantaAnadelValleZapotec,alanguagewheredifferencesinphonationconveyimportantlinguisticinformation.Resultswereinconclusiveastowhetherornotonegenderwascreakierorbreathierthantheotherbecauseofthesuccessofdifferentacousticmeasuresforthetwogenders.However, ContrastivephonationinSantaAnaDelValleZapotec Focus (high Range S.D. Range S.D. Range S.D. Range S.D. Range S.D. 1 to 5 1.98 1 to 6 2.36 8 to 12 1.87 7 to 11 2.00 14 to -1 to 2 1.91 0 to 3 2.03 -1 to 4 2.01 -1 to 0 1.12 -3 to 0 1.80 -3 to 0 2.12 -5 to 0 2.52 -10 to -8 1.43 -3 to 1 2.98 -5 to 0 2.90 Figure13AverageH1-A3valuesforbreathy,modal,andcreakyphonationsinveprosodicpositionsforSpeaker3(male).Therangeandstandarddeviationsarepresentedbelowthegraph. Figure14AverageF0forSpeaker3(male)inveprosodicpositions(isolationwithfocus,initial,isolation(notfocused),medial,andnalposition)atthreetimepoints(1,2,3)pervowel.thesuccessofH1-H2forfemalesandH1-A3formalessuggeststhatthereisadifferenceinthePRODUCTIONofphonationbetweenthegenders,thoughthereasonforthisdifferenceremainsunknown.Presumably,allhealthyhumansarecapableofproducingallpossiblearticulations;perhapsthegender-basedphonationdifferencesinSADVZaredeterminedbysociolinguisticfactors.However,Sodersten&Lindestad(1990)didshowthatfemalespeakers(ofEnglish) ChristinaM.Esposito Final (low F0) Range S.D. Range S.D. Range S.D. Range S.D. Range S.D. 1 to 4 1.87 0 to 5 2.17 10 to 19 3.03 10 to 15 2.41 19 to 24 1.67 -1 to 1 1.84 -2 to 1 2.31 1 to 5 2.00 0 to 6 3.12 4 to 9 2.34 -2 to 3 2.04 -2 to -1 1.03 -8 to -1 1.93 -3 to 0 1.95 -10 to 0 2.84 Figure15AverageH1-H2valuesforbreathy,modal,andcreakyphonationsinveprosodicpositionsforSpeaker4(female).Therangeandstandarddeviationsarepresentedbelowthegraph. Figure16AverageF0forSpeaker4(female)inveprosodicpositions(isolationwithfocus,initial,isolation(notfocused),medial,andnalposition)atthreetimepoints(1,2,3)pervowel. ContrastivephonationinSantaAnaDelValleZapotec Focus (high F0) Final (low F0) Range S.D. Range S.D. Range S.D. Range S.D. Range S.D. 0 to 6 2.10 2 to 7 1.83 9 to 13 1.31 9 to 13 1.21 19 to 23 1.93 0 to 3 1.86 1 to 3 1.90 0 to 3 1.21 0 to 2 1.18 -5 to 0 2.08 -8 to 0 3.01 -5 to 0 2.00 -11 to -8 1.24 -11 to -8 1.04 -8 to -14 2.00 Figure17AverageH1-H2valuesforbreathy,modal,andcreakyphonationsinveprosodicpositionsforSpeaker5(female).Therangeandstandarddeviationsarepresentedbelowthegraph. Figure18AverageF0forSpeaker5(female)inveprosodicpositions(isolationwithfocus,initial,isolation(notfocused),medial,andnalposition)atthreetimepoints(1,2,3)pervowel.weremorelikelytohaveincompleteglottalclosure.Thus,anotherpossibleexplanationforthegender-baseddifferencescouldbeaphysiologicalone,thoughdirectobservationsofthevocalfolds,thelarynxandthevalvesofthethroatareneededtoverifythishypothesis.Furthermore,theseresultsraisethequestionofwhetherornotSADVZlistenerscanperceivethedifferencebetweenH1-H2andH1-A3,andwhetherornotthesedifferencesareassociated ChristinaM.Espositowithgenderdifferences(i.e.doH1-H2differencessoundÔfeminineÕwhileH1-A3differencessoundÔmasculineÕ?).Aperceptionstudyinvestigatingtheseissueswouldbeanimportantfollow-uptothecurrentstudy.AdditionalÞndingsshowedthatF0,independentofposition,hadastrongeffectonphonation.WhenF0washigh(i.e.isolationandinitialposition),thethree-waycontrastinphonationwasminimized,withnon-modalphonationhavinganH1-H2/A3valuetypicallyassociatedwithmodalphonation.(WhiletheH1-H2/A3valuesathighF0indicateamodal-likephonation,itispossiblethatthecreakyphonationisnotactuallymoremodal,butratheratypeofharshorpressedvoice.)ThesamepatternisalsoseenwithlexicalF0;inSADVZmodalphonationisproducedwithhighF0sandnon-modalphonationsareproducedwithlowerF0s.Cross-linguistically,highF0,whichcanincreasevocalfoldtensionandlength,isassociatedwithmodalphonation(thoughF0isnotalwayscorrelatedwithphonation;seeLadefoged1973,Laver1980).ThisglottalconÞguration(i.e.increasedvocalfoldtensionandlength)istheoppositeofthetypicalconÞgurationforbreathyphonation,whichcouldexplainwhyspeakersareproducingphonemicallybreathyphonationswithH1-H2/A3valuestypicallyassociatedwithmodalphonationinenvironmentswithhighF0.Furthermore,creakyphonationcanalsoloseitscharacteristiccreakathighF0s.Creakyphonation(withfastclosureandlongandslowvocalfoldopening)isproducedwhenthearyepiglotticsphincterisengaged(Esling2005).Thisshortensthelengthoftheglottisandoftheepilaryngealstructuresovertheglottis.WhenF0ishigh,thelarynxwilllengthenlongitudinally,stretchingthevocalfoldsintheoppositedimensionofthetypicalconÞgurationforcreakyphonation.(Formoreinformationontherelationshipbetweenphonationtypeandpitch,seeEsling&HarrisTheresultsobtainedinthisstudyraiseaninterestingissueforÞeldworkersworkingonZapoteclanguages,andperhapslanguageswithphonationcontrastsingeneral.InSantaAnaDelValleZapotec,whenF0ishigh,thephonemicthree-waycontrastinphonationisnotwellpreserved.Thus,wordselicitedinisolationwithfocus,whichhaveahigherF0thansentence-medialposition,willnotshowanobviousphonationcontrast.Inordertoseeafullrangeofphonationcontrasts,itisimportanttoelicitdatathatdisplayafullrangeofF0s.Thecurrentstudyshowsthatthereisvariationinphonation,eveninlanguageswithcontrastivephonation.However,itremainstobeseenhowSADVZcompareswithotherlanguageswithcontrastivephonation.Inordertotrulyunderstandthenatureofthetypeofvariationdemonstratedinthisstudy,itisnecessarytoreplicatetheseresultswithotherlanguagesthatcontrastphonationandtoelicitdatainmorenaturaldiscoursecontexts.AcknowledgementsThisarticlederivesfromalargerresearchprojectguidedbyMatthewGordon,PatriciaKeating,andSun-AhJun.Ithankthemfortheirtimeandguidance.ThanksalsotomyZapotecconsultantsforprovidingthedatausedinthispaper.ReferencesAndruski,Jean&MarthaRatliff.2000.Phonationtypesinproductionofphonologicaltone:ThecaseofGreenMong.JournaloftheInternationalPhoneticAssociation30(1/2),37Ð61.Bickley,Corine.1982.Acousticanalysisandperceptionofbreathyvowels.SpeechCommunicationGroupWorkingPapers1,73Ð93.Cambridge,MA:MIT.Blankenship,Barbara.1997.Thetimecourseofbreathinessandlaryngealizationinvowels.Ph.D.dissertation,UniversityofCalifornia,LosAngeles.Dilley,Laura,StephanieShattuck-Hufnagel&M.Ostendorf.1996.Glottalizationofword-initialvowelsasafunctionofprosodicstructure.JournalofPhonetics24,423Ð444. ContrastivephonationinSantaAnaDelValleZapotecEdmondson,JeroldA.&JohnH.Esling.2006.Thevalvesofthethroatandtheirfunctioningintone,vocalregister,andstress:Laryngoscopiccasestudies.Phonology23(2),157Ð191.Epstein,Melissa.2002.VoicequalityandprosodyinEnglish.Ph.D.dissertation,UniversityofCalifornia,LosAngeles.Esling,JohnH.2005.Therearenobackvowels:Thelaryngealarticulatormodel.CanadianJournalof50,13Ð44.Esling,JohnH.&JimmyHarris.2005.Statesoftheglottis:Anarticulatoryphoneticmodelbasedonlaryngoscopicobservations.InWilliamJ.Hardcastle&JanetBeck(eds.),AÞgureofspeech:AFestschriftforJohnLaver,347Ð383.Mahwah,NJ:LawrenceErlbaum.Esposito,ChristinaM.2002.Pilotstudy:SantaAnadelValleZapotecPhonation.Ms.,UniversityofCalifornia,LosAngeles.Esposito,ChristinaM.2003.SantaAnadelValleZapotecphonation.MAthesis,UniversityofCalifornia,LosAngeles.Esposito,ChristinaM.2004a.SantaAnadelValleZapotecphonation.UCLAWorkingPapersinPhonetics103,71Ð105.Esposito,ChristinaM.2004b.SantaAnadelValleZapotecIntonation.Ms.,MacalesterCollege.Fischer-J¿rgensen,Eli.1967.Phoneticanalysisofbreathy(murmured)vowels.IndianLinguisticsFujimura,Osamu(ed.).1988.Vocalphysiology:Voiceproduction,mechanismsandfunctions,297Ð317.NewYork:RavenPress.Fujimura,Osamu&MinoruHirano(eds.).1995.Vocalfoldphysiology:Voicequalitycontrol.SanDiego,CA:SingularPublishingGroup.Gobl,Christopher.1988.Voicesourcedynamicsinconnectedspeech.1,123Ð159.Gordon,Matthew&PeterLadefoged.2001.Phonationtypes:Across-linguisticoverview.Journalof29,383Ð406.Gordon,Raymond,Jr.(ed.),2005.Ethnologue:Languagesoftheworld,15thedn.Dallas,TX:SILInternational.http://www.ethnologue.com/web.asp(March2009).Hagen,Astrid.1997.LinguisticfunctionsofglottalizationsandtheirlanguagespeciÞcuseinEnglishand.Ph.D.dissertation,Friedrich-Alexander-UniversitatErlangen-Nurnberg&MIT.Hanson,Helen.1997.Glottalcharacteristicsoffemalespeakers:Acousticcorrelates.JournaloftheAcousticalSocietyofAmerica101(1),466Ð481.Hanson,Helen&ErikaChuang.1999.Glottalcharacteristicsofmalespeakers:Acousticcorrelatesandcomparisonwithfemaledata.JournaloftheAcousticalSocietyofAmerica106(2),1064ÐHenton,Caroline&R.AnthonyBladon.1985.Breathinessinnormalfemalespeech:InefÞciencyversusLanguageandCommunication5(3),221Ð227.Holmberg,EvaB.,RogerE.Hillman,JosephPerkell,PeterGuiod&SusanL.Goldman.1995.Comparisonsamongaerodynamic,electroglottographic,andacousticspectralmeasuresoffemalevoice.JournalofSpeech,Language,andHearingResearch38,1212Ð1223.Huffman,Marie.1987.MeasuresofphonationtypeinHmong.JournaloftheAcousticalSocietyofAmerica81(2),495Ð504.Kirk,PaulL.,JennyLadefoged&PeterLadefoged.1993.Quantifyingacousticpropertiesofmodal,breathy,andcreakyvowelsinJalapaMazatec.InAnthonyMattina&TimothyMontler(eds.),IndianlinguisticsandethnographyinhonorofLaurenceC.Thompson,435Ð450.Missoula,MT:UniverstiyofMontanaPress.Klatt,Dennis&LauraKlatt.1990.Analysis,synthesisandperceptionofvoicequalityvariationsamongfemaleandmaletalkers.JournaloftheAcousticalSocietyofAmerica87(2),820Ð857.Kreiman,Jody.1982.Perceptionofsentencesandparagraphboundariesinnaturalconversation.JournalofPhonetics10,163Ð175.Ladefoged,Peter.1971Preliminariestolinguisticphonetics.Chicago:UniversityofChicagoPress.Ladefoged,Peter.1983.Thelinguisticuseofdifferentphonationtypes.InDianeBless&JamesAbbsVocalfoldphysiology:Contemporaryresearchandclinicalissues,351Ð360.SanDiego,CA:College-HillPress. ChristinaM.EspositoLadefoged,Peter,IanMaddieson&MichaelJackson.1988.Investigatingphonationtypesindifferentlanguages.InFujimura(ed.),297Ð317.Laver,John.1981.Phoneticdescriptionofvoicequality.Cambridge:CambridgeUniversityPress.Lehiste,Ilse.1975.Thephoneticstructureofparagraphs.InAntonieCohen&SiboutNooteboom(eds.),Structureandprocessinspeechperception,195Ð203.Heidelberg&NewYork:Springer.Maddieson,Ian&SusanHess.1987.TheeffectofF0onlinguisticsuseofphonationtypes.UCLAWorkingPapersinPhonetics67,112Ð118.Munro,Pamela&FelipeLopez,withOliviaMendez,RodrigoGarcia&MichaelGalant.1999.DiÕsyonaaryX:teeÕnDiizhSahSannLuÕuc(SanLucasQuiavinõDictionary/DiccionarioZapotecodeSanLucas.LosAngeles,CA:ChicanoStudiesResearchCenterPublications.Ogden,Richard.2003.Voicequalityasaresourceforthemanagementofturn-takinginFinnishtalk-in-15thInternationalConferenceofPhoneticSciencesXV),Barcelona,123Ð126.Ohala,John.1973.Thephysiologyoftone.SouthernCaliforniaOccasionalPapersinLinguistics1,1Ð14.Pierrehumbert,Janet.1995.Prosodiceffectsonglottalallophones.InFujimura&Hirano(eds.),39Ð60.Pierrehumbert,Janet&DavidTalkin.1992.Lenitionof//andglottalstop.InGerardJ.Docherty&D.RobertLadd(eds.),PapersinLaboratoryPhonologyII:Gesture,segment,prosody,90Ð117.Cambridge:CambridgeUniversityPress.Redi,Laura&StephanieShattuck-Hufnagel.2001.Variationintherealizationofglottalizationinnormalspeakers.JournalofPhonetics29,407Ð429.Shue,Yen,PatriciaKeating&ChadVicenik.2009.VoiceSauce:Aprogramforvoiceanalysis.JournaloftheAcousticalSocietyAmerica124(4),2221.Silverman,Daniel,BarbaraBlankenship,PaulKirk&PeterLadefoged.1995.PhoneticstructuresofJalapaAnthropologicalLinguistics37,70Ð88.oderstenMaria&Per-AkeLindestad.1990.Glottalclosureandperceivedbreathinessduringphonationinnormallyspeakingsubjects.JournalofSpeechandHearingResearch33(3),601Ð611.Stevens,Kenneth.1977.Physicsoflaryngealbehaviorandlarynxmodes.34,264Ð279.Stevens,Kenneth.1988.Modesofvocalfoldvibrationbasedonatwo-sectionmodel.InFujimura(ed.),Stevens,Kenneth&HelenHanson.1995.ClassiÞcationofglottalvibrationfromacousticmeasurements.InFujimura&Hirano(eds.),147Ð170.Todaka,Yuichi.1993.Across-languagestudyofvoicequality:BilingualJapaneseandAmericanEnglishspeakers.Ph.D.dissertation,UniversityofCalifornia,LosAngeles. VariationincontrastivephonationinSantaAnaDelValleZapotecChristinaM.EspositoDepartmentofLinguistics,MacalesterCollegeesposito@macalester.eduThepresentstudysetsouttoinvestigatevariationduetogender,F0,and/orprosodicpositioninSantaAnadelValleZapotec(Oto-Manguean),alanguagewithphonemicallybreathy, Ladefoged(1971)proposedacontinuumofphonationtypescharacterizedbytheirdegreeofglottal