Proceedings Fonetik 2009 - Institutionen för lingvistik

Proceedings, FONETIK 2009, Dept. of Linguistics, Stockholm Universityity. If it is found that children indeed recognizetheir recorded voice as their own, this may haveimportant implications for the use of recordingsin speech and language intervention.PurposeThe purpose of this study is to explore children’sability to recognize recordings of theirown voice as their own, and whether this abilityvaries depending on the age of the child and thetime between the recording and the listening.The research questions are:1. Are children with normal hearing ableto recognize their own recorded voiceas their own, and identify it when presentedtogether with 3 other childvoices?2. Will this ability be affected by the timespan between recording and listening?3. Will the performance be affected by theage of the child?as references. None of the reference childrenwere known to the children in the test groups.Recording/Identification procedureA computer program was used to present thewords in the scripts in random order, and foreach word1. Play a reference voice (adult) that readsa target word, while displaying a picturethat illustrates the word.2. Record the subject’s production of thesame word (with the possibility of listeningto the recording and re-recordinguntil both child and experimenter aresatisfied).3. Play the subject’s production and 3 referencechildren’s productions of thesame word, in random order, letting thesubject select one of these as his/herown. (See Figure 1.)It is hypothesized that the older children willperform better than the younger children, andthat both age groups will perform better whenlistening immediately after the recording thanwhen listening 1-2 weeks after the recording.MethodParticipants45 children with Swedish as their mothertongue, and with no known hearing problemsand with no previous history of speech and languageproblems or therapy were invited to participate.The children were divided into two agegroups, with 27 children aged 4-5 years (rangingfrom 4;3 to 5;11, mean age 5;3) in theyounger group and 18 children aged 7-8 years(ranging from 7;3 to 8;9, mean age 8;0) in theolder group. Only children whose parents didnot know of or suspect any hearing or languageproblems in the child were invited. All childrenwere recruited from pre-schools in Stockholm.MaterialA recording script of 24 words was constructed(see Appendix). The words in the script all beganwith /tV/ or /kV/, and all had primary stresson the first syllable.Three 6-year old children (two girls and oneboy, included by the same criteria as the childrenparticipating in the study) were recordedFigure 1: The listening/identification setup.In both test sessions, the children were fittedwith a headset and the experimenter with headphonesto supervise the recordings. The childrenwere instructed to select the character theybelieved represented their own voice by pointingat the screen; the actual selection was managedby the experimenter by mouse clicking.The children were given two introductory trainingitems, to assure understanding of the task.In the first test session, the children performedboth the recording and the voice identificationtask, i.e. step 1-3. For the recordings,all children were instructed to speak with theirnormal voice, and utterances were re-recordeduntil both child and experimenter were satisfied.In the second test session, after a period of1-2 weeks, the children performed only theidentification task, i.e. step 3. Apart from generalencouragement, the experimenter provided137