Proceedings Fonetik 2009 - Institutionen för lingvistik

More documents

Recommendations

Info

Proceedings, FONETIK 2009, Dept. of Linguistics, Stockholm UniversityEarwitnesses: The effect of voice differences in identificationaccuracy and the realism in confidence judgmentsElisabeth Zetterholm 1 , Farhan Sarwar 2 and Carl Martin Allwood 31 Centre for Languages and Literature, Lund University2 Department of Psychology, Lund University3 Department of Psychology, University of GothenburgAbstractIndividual characteristic features in voice andspeech are important in earwitness identification.A target-absent lineup with six foils wasused to analyze the influence of voice andspeech features on recognition. The participants’response for two voice foils were particularlysuccessful in the sense that they weremost often rejected. These voice foils werecharacterized by the features’ articulation rateand pitch in relation to the target voice. For thesame two foils the participants as a collectivealso showed marked underconfidence and especiallygood ability to separate correct andincorrect identifications by means of their confidencejudgments for their answers to the identificationquestion. For the other four foils theparticipants showed very poor ability to separatecorrect from incorrect identification answersby means of their confidence judgments.IntroductionThis study focuses on the effect of some voiceand speech features on the accuracy and the realismof confidence in earwitnesses’ identifications.More specifically, the study analyzes theinfluence of characteristic features in thespeech and voices of the target speaker and thefoils in a target-absent lineup on identificationresponses and the realism in the confidence thatthe participants feel for these responses. Thistheme has obvious relevance for forensic contexts.Previous research with voice parades hasoften consisted of speech samples from laboratoryspeech, which is not spontaneous (Cook &Wilding, 1997; Nolan, 2003). In spontaneousspeech in interaction with others, the assumptionis that the speakers might use anotherspeaking style compared with laboratoryspeech. In forensic research spontaneousspeech is of more interest since that is a morerealistic situation.Sex, age and dialect seem to be strong anddominant features in earwitness identification(Clopper et al, 2004; Eriksson et al., 2008; Lasset al., 1976; Walden et al., 1978). In these studies,there is nothing about how the witness’confidence and its’ realism is influenced bythese features.The study presented in this paper focuses onthe influence of differences and similarities invoice and speech between a target voice and sixfoils in a lineup. A week passed between theoriginal presentation of the target speaker (atfor example the crime event) and the lineup,which means that there is also a memory effectfor the listeners participating in the lineup.Spontaneous speech is used in all recordingsand only male native Swedish speakers.Confidence and realism in confidenceIn this study, a participant’s confidence in hisor her response to a specific voice in a voiceparade with respect to if the voice belongs tothe target or not, relates to whether this responseis correct or not. Confidence judgmentsare said to be realistic when they match the correctness(accuracy) of the identification responses.Various aspects of realism can bemeasured (Yates, 1994). For example, the over-/underconfidence measure indicates whetherthe participant’s (or group’s) level of confidencematches the level of the accuracy of theresponses made. It is more concretely computedas: Over-/underconfidence = (The meanconfidence) minus (The mean accuracy).Another aspect of the realism is measuredby the slope measure. This measure concerns aparticipant’s (or group’s) ability, by means ofone’s confidence judgments, to, as clearly aspossible, separate correct from incorrect judgments.This measure is computed as: Slope =(The mean confidence for correct judgments)minus (The mean confidence for incorrectjudgments). The relation between a participant’slevel of confidence for a voice with a180
Proceedings, FONETIK 2009, Dept. of Linguistics, Stockholm Universityspecific characteristic compared with the participant’sconfidence for the other voices in thelineup might indicate how important that characteristicis for the participant’s judgment. In aforensic context the level of realism indicateshow useful a participant’s confidence judgmentsare in relation to targets’ and foils’voices with specific features.MethodParticipants100 participants took part in the experiment.The mean age was 27 years. There were 42males and 58 females. 15 participants had anothermother tongue than Swedish, but they allspeak and understand Swedish. Four of themarrived in Sweden as teenagers or later. Fiveparticipants reported minor impaired hearingand four participants reported minor speechimpediment, one of them stuttering.MaterialsThe dialogue of two male speakers was recorded.They played the role of two burglarsplanning to break into a house. This recordingwas about 2 minutes long and was used as thefamiliarization passage, that is, as the originalexperienced event later witnessed about. Thespeakers were 27 and 22 year old respectivelywhen recorded and both speak with a Scaniandialect. The 22 years old speaker is the targetand he speaks most of the time in the presentedpassage.The lineup in this study used recordings ofsix male speakers. It was spontaneous speechrecorded as a dialogue with another malespeaker. This male speaker was the same in allrecordings and that was an advantage since hewas able to direct the conversation. They alltalked about the same topic, and they all hadsome kind of relation to it since it was an ordinarysituation. As a starting point, to get differentpoints of view on the subject talked aboutand as a basis for their discussion, they all readan article from a newspaper. It had nothing todo with forensics. The recordings used in thelineups were each about 25 sec long and only apart of the original recordings, that is, the maleconversation partner is not audible in the lineups.All the six male speakers have a Scaniandialect with a characteristic uvular /r/ and aslightly diphthongization. They were chosenfor this study because (as described in more detailbelow) they share, or do not share, differentfeatures with the target speaker. It is featuressuch as pitch, articulation rate, speaking styleand overall tempo and voice quality.The target speaker has a mean F0 (meanfundamental frequency) of 107 Hz, see Table 1.The speech tempo is high overall and he has analmost forced speaking style with a lot of hesitationsounds and repetition of syllables whenhe is excited in the familiarization passage. Theacoustic analysis confirms a high articulationrate.Foil 1 and 2 are quite close in their speechand voices in the auditory analysis. Both speakwith a slightly creaky voice quality, althoughfoil 2 has a higher mean F0. Their articulationrate is quite high and close to the targetspeaker. Foil 3 and 6 speak with a slowerspeech tempo and a low and a high pitch respectivelyis audible. In the acoustic analysis itis also obvious that both foil 3 and 6 have anarticulation rate which is lower than the targetspeaker. Foil 4 is the speaker who is closest tothe target speaker concerning pitch and speakingstyle. He speaks with a forced, almost stutteringvoice when he is keen to explain something.His articulation rate is high and he alsouses a lot of hesitation sounds and filled pauses.Foil 5 has quite a high articulation rate, in similarityto the target speaker, but he has a higherpitch and his dialect is not as close to the targetspeaker as the other foils.All the speakers, including the targetspeaker, are almost the same age, see Table 1.The results of the acoustic measurements ofmean fundamental frequency (F0) and std.dev.are also shown in the table. The perceptualauditory impression concerning the pitch isconfirmed in the acoustic analysis.Table 1. Age, F0 mean and standard deviations(SDs) for the target speaker and the six foilsAge F0, mean SDs.target 22 107 Hz 16 Hzfoil 1 23 101 Hz 26 Hzfoil 2 21 124 Hz 28 Hzfoil 3 23 88 Hz 15 Hzfoil 4 23 109 Hz 19 Hzfoil 5 23 126 Hz 21 Hzfoil 6 25 121 Hz 17 HzProcedureThe experimental sessions took place in classesat the University of Lund and in classes with181
Page 1 and 2:
Department of LinguisticsProceeding
Page 3 and 4:
Proceedings, FONETIK 2009, Dept. of
Page 5 and 6:
Page 7 and 8:
Page 9 and 10:
Page 11 and 12:
Page 13 and 14:
Page 15 and 16:
Page 17 and 18:
Page 19 and 20:
Page 21 and 22:
Page 23 and 24:
Page 25 and 26:
Page 27 and 28:
Page 29 and 30:
Page 31 and 32:
Page 33 and 34:
Page 35 and 36:
Page 37 and 38:
Page 39 and 40:
Page 41 and 42:
Proceedings, FOETIK 2009, Dept. of
Page 43 and 44:
Page 45 and 46:
Page 47 and 48:
Page 49 and 50:
Page 51 and 52:
Page 53 and 54:
Page 55 and 56:
Page 57 and 58:
Page 59 and 60:
Page 61 and 62:
Page 63 and 64:
Page 65 and 66:
Page 67 and 68:
Page 69 and 70:
Page 71 and 72:
Page 73 and 74:
Page 75 and 76:
Page 77 and 78:
Page 79 and 80:
Page 81 and 82:
Page 83 and 84:
Page 85 and 86:
Page 87 and 88:
Page 89 and 90:
Page 91 and 92:
Page 93 and 94:
Page 95 and 96:
Page 97 and 98:
Page 99 and 100:
Page 101 and 102:
Page 103 and 104:
Page 105 and 106:
Page 107 and 108:
Page 109 and 110:
Page 111 and 112:
Page 113 and 114:
Page 115 and 116:
Page 117 and 118:
Page 119 and 120:
Page 121 and 122:
Page 123 and 124:
Page 125 and 126:
Page 127 and 128:
Proceedings, FOETIK 2009, Dept. of
Page 129 and 130: Proceedings, FOETIK 2009, Dept. of
Page 137 and 138: Proceedings, FONETIK 2009, Dept. of
Page 179: Proceedings, FONETIK 2009, Dept. of
Page 227: Department of LinguisticsPhonetics
show all

Proceedings Fonetik 2009 - Institutionen för lingvistik

You also want an ePaper? Increase the reach of your titles

Delete template?

Save as template?