PSY369 Psycholinguistics LanguageComprehensionSpeechrecognition Differentfeaturesthanvisual Visualwordrecognition SpeechPerception SomeparallelinputOrthographyLettersCleardelineationDifficulttolearn SerialinputPhonetics PhonologyAcousticfeaturesUsuallynodelineation Easy tolearn Differentproblemstoo ChestJewWadeAimInItJustyouwaitaminuteDelightsHavenDimeDaylightSavingsTimeCanoesHeWadIceHeCanYouSeeWhatISee FreeQuaintlyAsQuestShunsFrequentlyAskedQuestions HardProblemsinSpeechPerception Linearity paralleltransmission AcousticfeaturesoftenspreadthemselvesoutoverothersoundsWheredoesshowstartandmoneyend Waveform Demo sandinfo Showmethemoney HardProblemsinSpeechPerception Segmentationproblem Unlikevisualinput theacousticinputisnotphysicallysegmentedIllusionofsilence Therearenosilentgapsinthewaveform eventhoughwemay hear some IoweyouaYo Yo HardProblemsinSpeechPerception Segmentationproblem Unlikevisualinput theacousticinputisnotphysicallysegmentedIllusionofsilence Therearenosilentgapsinthewaveform eventhoughwemay hear some HardProblemsinSpeechPerception Segmentationproblem Unlikevisualinput theacousticinputisnotphysicallysegmentedHerethesilencethatweseeintheacousticsisn tperceivedasagapintheword HardProblemsinSpeechPerception LackofInvariance OnephonemeshouldhaveaonewaveformThisisnotthecase The i ee in money and me aredifferent Waveform Showmethemoney HardProblemsinSpeechPerception LackofInvariance OnephonemeshouldhaveaonewaveformAnotherexample Hereisthephoneme d followedbydifferentvowels HardProblemsinSpeechPerception LackofInvariance OnephonemeshouldhaveaonewaveformAndanother Thephrasehasfive t phonemes buttherearenot5identicalsweepsinthespectrogramTherearen tinvariantcuesforphoneticsegmentsAlthoughthesearchcontinues Peterbutteredtheburnttoast HardProblemsinSpeechPerception Co articulation theinfluenceofthearticulation pronunciation ofonephonemeonthatofanotherphoneme Essentially producingmorethanonespeechsoundatonceMaybehelpfulbecauseitallowssomeparalleltransmissionofinformation possiblyhelpingpredictwhat scomingnext Eachsoundpartiallyshapedbysoundsbefore afteritkeelvskillvscool kil vs kIl vs kul IPAcharacters placeofarticulationandroundingonthekdifferalotdifferentversionsof thesamesound indifferentcontextsfromdifferentspeakers ThisiswhatallowsustotalksofastMaybehelpfulbecauseitallowssomeparalleltransmissionofinformation possiblyhelpingpredictwhat scomingnext HardProblemsinSpeechPerception TradingrelationsMostphoneticdistinctionshavemorethanoneacousticcueasaresultoftheparticulararticulatorygesturethatgivesthedistinction Voice onset time VOT EnergyinburstOnsetfrequencyofthefirstformantPlacementinsyllablee g slit split the p reliesonsilenceandrisingformant differentmixturesofthesecanresultinthesameperceptionPerceptionmustestablishsome trade off betweenthedifferentcues HardProblemsinSpeechPerception ManyfactorsthatmaybeimportantAcousticInformationVisualinformationProsodicinformationSemanticcontextSyntacticstructure UsingVisualinformation McGurkeffectMcGurkeffect2 TheMcGurkeffect McGurkandMacDonald 1976 Showedpeopleavideowheretheaudioandthevideodon tmatchThink dubbedmovie Visual ga withauditory ba oftenhear da ImplicationsPhonemeperceptionisanactiveprocessInfluencedbybothaudioandvisualinformation Beyondthesegment Prosodicfactors suprasegmentals English Speechisdividedintophrases Everyphrasehasafocus WordstressismeaningfulinEnglish Stressedsyllablesarealignedinafairlyregularrhythm whileunstressedsyllablestakeverylittletime Anextendedflatorlow risingintonationattheendofaphrasecanindicatethataspeakerintendstocontinuetospeak Afallingintonationsoundsmorefinal Beyondthesegment Prosodicfactors suprasegmentals StressEmphasisonsyllablesinsentencesOnmeaning blackbird versus blackbird Top downeffectsonperceptionBetteranticipationofupcomingsegmentswhensyllableisstressedRateSpeedofarticulation Fastertalking shortervowels shorterVOTNormalization takingthespeaker srateintoaccountIntonationUseofpitchtosignifydifferentmeaningsacrosssentences Top downeffectsonSpeechPerception SentencecontexteffectsExcisedspeechPhonemerestorationeffectSentencecontexteffects ExcisedSpeech SyntacticandsemanticcuescanhelpPollack Pickett 1964 Task Recordedconversationsandexcisedindividualwords PresentedthewordstolistenersforidentificationWithincontextOutofcontextResults Wordsoutofcontextwereonlyrecognized47 oftime identificationwasgreatlyimprovedwithcontextSuggeststhatclarityinspeechreflectsprocessing top downaswellasbottom up Phonemerestorationeffect Task Listentoasentencewhichcontainedawordfromwhichaphonemewasdeletedandreplacedwithanothernoise e g acough Thestategovernorsmetwiththeirrespectivelegi laturesconveninginthecapitalcity s deletedandreplacedwithacough Clickhereforademoandadditionalinformation Warren 1970 Results Participantsheardthewordnormally despitethemissingphonemeUsuallyfailedtoidentifywhichphonemewasmissingInterpretation Wecanusetop downknowledgeto fillin themissinginformation Phonemerestorationeffect WarrenandWarren 1970 Whatifthemissingpho