Auditory sensitivity in preschool children

Auditory sensitivity in preschool children BruceA. Schneider, SandraE. Trehub,BarbaraA. Morrongiello, andLeighA. Thorpe Department ofPsychology, E•ndaleCollege, University of Toronto, Mississauga, Ontario,CanadaLSL IC6

(Received9 January1985;acceptedfor publication25 October1985)

Thresholds for octave-band noises withcenterfrequencies of 0.4, 1,2, 4, and10kHz, and1/3-

octave-band noises withcehter frequencies of 10and20kHz,wereobtained fromchildren 3-5 yearsof ageandfroma comparison groupof adults.Thresholds for all frequencies decreased between 3 and5 yearsofage.Thresholds decreased furtherbetween 5 yearsof ageandadulthood, exceptfor the20-kHz stimulus,for whichchildrenhadlowerthresholds thanadults.These resultsarediscussed in termsof possible age-related changes in themechanical properties of the car andin the efficiency of neuralcoding. PACS numbers:43.66.Cb,43.66.Sr [DW]

INTRODUCTION

groupof adults.We adoptedthe test procedures usedby Trehub et aL (1980) and Schneideret aL (1980) with in-

In recentyears,therehasbeenincreasing interest in the specification ofauditory sensitivity in infancy(EiseleetaL, 1975;Weir, 1979;Schneider etaL, 1980;TrehubetaL, 1980; Bergand Smith, 1983;Sinottet aL, 1983;Thompsonand Wilson,1984).The generalpicturethat hasemergedis one of diminishedsensitivityfor infants relative to adults, althoughthe extentof the differenceremainsunresolved(Aslinet aL, 1983;WilsonandThompson,1984).Thereis also someindication of developmental changes in thepatternof sensitivity. For example,adult-infantthreshold differences seemto begreatest at thelowfrequencies andsmallest at the highfrequencies (Schneider etaL, 1980;TrehubetaL, 1980; Sinnott et aL, 1983}.

Thereis evidenceof continuingimprovementwell into theschoolyears(JordanandEagles,1963;HartleyandSicgentbaler,1964;Abromovitz,1971;Leninbanet aL, 1971; Fior, 1972;RocheetaL, 1978;Yost, 1978;MaxonandHochberg,1982)butlittleconsensus abouttheageat whichhear-

ing is mostsensitive or whenit reaches adultlevel.Fior (1972) suggests that "hearingmaturity"is reachedat approximately 10-13yearsof age.DatafromLeninbanet al. ( 1971) and Maxon and Hochberg(1982) are more or less consistentwith this view but those of Roche et aL ( ! 978)

indicatecontinuingincreases in sensitivitybetween12 and 17 yearsof age. The long-standing interestin auditorysensitivityin school-aged childrenandthecurrentburgeoning interestin infant sensitivitystandin sharpcontrastto the relativepaucityof research focused specifically onthepreschool period. Althoughit isclearthat preschool childrenhavehigherabsolutethresholdsthan school-aged children (Harticy and Siegenthaler,1964; Fior, 1972; Maxon and Hochberg, 1982), it is unclearhowsensitivitychangeswith ageduring the preschoolperiod. The purposeofthe presentinvestigation wasto chartthe developmentalcourseof auditory sensitivityin the preschoolperiod.Specifically, wesoughtto establish thresholds for narrow-band noiseswith centerfrequencies between0.4 and 10 kHz for children3-5 yearsof ageanda comparison 447

J. Acoust.Soc.Am. 79 (2), February1986

fants6-24 monthsof agein orderto facilitatecomparisons overan extendedagerange. I. METHOD

A. Subiceta

Childparticipants wererecruitedfromlocalpreschools, parentgroups,and from letterssent to nearbyfamilies. Adult participantswere studentswho volunteeredin response to postednotices. Thercwere251childrenin thefinal sample,53 at 3 yearsof age(36 months___ I month),51at 4 yearsof age(48 months-t- I month),and44 at 5 yearsof age (60 months-I- 1 month). All of thesechildrenwerehealthy, bornat term, had no historyof ear infections,and werefree of coldsat thetimeof testing.A comparison sampleof coldfreeadultsconsisted of 43 individuals,17-25 yearsof age, with a meanageof 20 years. B. Apparatusand stimuli

The experimentwas controlledby a microcomputer (CommodorePET, model2001), whichoperatedtheequipmentthrougha custom-builtinterface.The outputof a white noisegenerator(General Radio, model 1381) was filtered by a programmableBruel and Kjaer bandpassfilter (model 1617). The filter was set for octave bandwidths centered at 0.4, 1, 2, 4, and 10 kHz and for l/3-octave bandwidthscen-

teredat 10and20 kHz. Octavebandswerechosenprincipally because comparison datawereavailablefor youngerchildren (Trehub et aL, 1980) and for adults (Robinson and

Whittle,1964).Whilenarrowing thebandwidth wouldprovidebetterfrequency specificity, it wouldalsoincrease amplitudevariationin the soundfield (Dillon and Walker, 1982). Thus octavebandsminimizedamplitudevariation, therebyincreasingthe precisionof the thresholdmeasures,

whilestill retainingadequatestimulusspecificity. For frequencies of 10kHz andhigher,1/3-octavebands providea sufficient degreeof amplitudestability.Moreover, the presentationof octavebandscenteredat 20 kHz was

0001-4966/86/020447-06500.80

¸ 1986 AcousticalSocietyof Arner/ca

447

TABLE I. Testintensitylevelsfor eachstimulus andagegroup. Frequency (kHz) Bandwidth Agegroup 0.4

octave

preschool adult

I

octave

2

octave

octave

10

1/3 octave

20

1/3 octave

21

31

I

11

21

12

22

32 22

2

2

12

10

20

30

0

10

20

-- 8

12

22

32

2

12

22

5

15

25

adult

--

preschool

-- 3

?

17

0 -- 10

pre•hool adult

10

1!

preschool adult

octave

I -- 9

preschool adult

4

Intensity(dB SPL)

2 -- 8

preschool

adult

-- 13

--3

2?

7

17

preschool

17

27

37

47

adult

27

37

47

57

precludedby thepresentapparatus.At 10kHz, bothoctave and 1/3-octavebandswereemployedto providecontinuity with the lowerandupperfrequencies, respectively. The rate of decrease in energyon eithersideof thebandwas30 dB per

the linearscaleweretakenwith a 0.5-in. microphonefor all stimuliexceptthe20-kHz signal,for whicha 0.25-in.microphonewasused.The testlevelsfor eachagegroupandstimulusareshownin TableI. Signallevelsfor mostfrequencies weredeterminedon the basisof a pilot study.In the caseof the 20-kHz stimulus, audible distortion occurred at about 62

dB; asa result,we selectedintensitiesof 57 dB and belowat thisfrequency.The ambientnoiselevelwas 16 dBA. The loudspeakers wereplacedin an IndustrialAcoustics sound-attenuating chamber (double-wall,measuring 3X2.8X2m), 1.8 m from the centerof the listener'schair,

whichoccupied onecornerof the room.A chairfor the experimenter waslocatedin thecorneropposite the listener's chair.Eachloudspeaker wasat a 45' angleto the listener's left and right. On top of eachloudspeaker was a fourchamber,smokedPlexiglas boxwithfourdifferentmechanical toysthat servedto reinforcecorrectresponses. Adjacent to eachloudspeaker wasa portablecolortelevision set(Sony Trinitron, model KV-1911), which was also used for rein-

forcement;the viewingscreenwasblackedout duringperiods of nonreinforcement.

octave.The outputof thefilterwasroutedto an impedanceC. Procedure matchingamplifierand,from there,the signalwassplitand fed to two programmable attenuators.The outputof each Duringthetestsession, S-year-old childrenandadults attenuatorfedan electronicswitchthat shapedthe signalto wereseatedin a testchairequippedwith a pushbuttonon each arm. These listeners were instructed to indicate their havea riseand decaytime of 40 ms,which eliminatedclicks at soundonsetand offset.The output of eachswitchwas judgmentof thelocationof thesignalby pressing thebutton routed to a preamplifier(SAE, model P101) by meansof onthecorresponding side;theirresponses wererecordedauthechild'smothersatin relay circuitry. Each output channelof the preamplifier tomatically.For 3- and4-year-olds, droveonechannelof a stereoamplifier(SAE, model2600). thetestchairwith thechildonherlapfacingtheexperimenter. These children were instructed to look at the sound The outputof eachchannelof the amplifierdrovean ESSHeil (modelAMTIAM) loudspeaker. source.Specifically, all childrenweretoldto "try to findthe Sound-pressure levelswerecalibratedwith a Brueland sound"andthat,if theydidso,theywouldgetto.see oneof Kjaer impulseprecisionsound-level meter(model2204) at thetoys.The requiredresponse for 3- and4-year-olds wasa theapproximate locationof thelistener'shead.Readingson headturnof 45øor greater,whichtheexperimenter recorded

n-

FIG. 1. Percentageof correctresponses asa function of decibel level of seven test fre-

i-

Z

quencies for children3-5 yearsof ageand for adult•(A).

n-

LU ea

RELATIVE

SOUND

PRESSURE

LEVEL

IN 10d8

STEPS I

448

J. Acoust.Soc. Am.,VoL79, No. 2, February1986

Schneiderota/.: Hearingin preschoolchildren

448

bymeans of a hand-held button-press box.Duringthetest. mutationsof the four intensitylevelswere generated(for ) sothatall fourintensitylevelswererepresentsession, theparent,if present, andtheexperimenter wore eachsubject signalassociated headphones withcontinuous whitenoiseto prevent them ed in a blockof fourtrials.The bandpass with eachof theseintensitylevelswasrandomlyassigned fromdetecting thelocusofthetestsignal. withtheconstraint thateachbandThetaskwasatwo-alternative forced-choice procedure. (again,foreachsubject) Theexperimenter pressed a buttonto initiatea trialonly passsignalcouldappearonlyonceat a givenintensitylevel. whenthe listenerwaslookingdirectlyaheadat her. The signalremained onuntilthelistenerresponded, eitherbya headturn ofatleast45øhathecaseoftheyounger children, or by a buttonpress,in thecaseof theolderchildrenandadults.

Whenthelistenerdid not respondwithha4 or 5 s, heor she

Thus each block of four trials contained one of the seven

stimuliat its highestintensitylevel.Assignmentof signalto loudspeaker locationwasrandomizedwith the constraints that anequalnumberof signalswouldappearontheleft and right loudspeakers and that no morethan threesuccessive trialscouldinvolvesignalpresentation from the sameside.

wastoldthatthesoundwasonandwasencouraged tolisten carefullyandguess the soundlocation.Correctresponses resultedin reinforcement, eitherthepresentation of an ani- II. matedtoy or TV cartoonsegment,for 4 s. Incorrectresponses resulted in a 4-stime-outperiodduringwhichtrials werenotpresented. The mechanical toyswereusedduring thefirstsession witheachsubject,andtheTV cartoons, duringthesecondsession. Listenerswhoparticipated in a third testsession couldselecteithertypeof reinforcement for that

RESULTS

The percentage of correctlocalizationresponses averagedacross subjects asa function ofsound-pressure levelfor eachof theseventeststimuliisshownin Fig. 1.The parameteroftheseven psyehometric functions at eachfrequency is the ageof the listeners, andeachpointon the functionsis based on a minimum of 100 trials. For all but the 20-kHz

session.

stimulus,the psychometfic functionsshiftto the left as a Duringtheinitialtrainingphase, listeners werepresent- functionof age,indicatingincreasing sensitivity with age.

ed with a 4-kHz octave-bandnoiseat 70 dB, the locationof

whichalternated between loudspeakers. Thepurpose of this alternation wasto preventperseverafive responding in the youngest children.Thechildrenwererequiredto makefour successive correctresponses at 70 dB, at whichtimethe intensitywasreducedto 60 dB; they wererequiredto make fourfurthersuccessive correctresponses at thisintensity level.Thetrainingefitefionforadultsrequiredonlytwosuccessivecorrectresponses at eachtrainingintensity.In subsequentsessions, childrenreceivedthis abbreviated training procedureandadultsreceivedno training.

Thetestphase, whichconsisted of28trials,immixiiately followed thetrainingphase. Eachtrial consisted of a single instanceof oneof the four intensitylevelsfor eachof the sevenbandpass noises.To minimizethe possibility that a listenermightencounter severalsubthreshold trialsin a row, thestimuliwererandomized asfollows.Sevenrandomper-

','•.

ß

I kHz \

o 2kHz

-,•

•

4 kHz

The amountof the shift is, however,frequencydependent, withthesmallestshiftoccurringto the 10-kHzoctave-band

noise. Thepsychometric functions for20kHz followa rather differentcourse. Theyshiftto theleftbetween3 and5 years of age,indicatingincreasing sensitivityacrossthis period. The functionfor adults,however,isto the fight of the functionsfor children,indicatingthat adultsare lesssensitive than childrenin this frequencyregion. Thesepsychometficfunctionswere usedto determine thresholdvaluesfor eachof the stimuliat eachagetested. Thresholdwas definedas the intensitycorresponding to

65% correct, • andwasdetermined by linearinterpolation between thetwointensities oneithersideof thispoint.In the twoinstances in whichperformance did not fall below65% ( 4 kHz for5-year-olds andadults),the65% pointwasdeterminedby meansof linearextrapolationfrom the two lowest intensityvalues.Figure 2 plotschildren'sthresholdsas a

10kHz FIG.

2. Thresholds for five octave-band

noisesasa functionof age.Also plotted are the data ofTrehub et al. (1980) for

infants0.5, l, and 1.5 yearsof age.

1234•12345123451234512345

AGE

449

IN YEARS

J. Acoust. Soc.Am.,Vol.79, No.2, February 1986

Schneider otat: Heatinginpreschool children

449

functionof agefor the octavenoisebands,togetherwith comparable thresholds fromTrehubetaL (1980) for infants 0.5, 1, and 1.5 yearsof age.Straight linesfitted to the data

pointsusingthe methodof leastsquares providea reasonably goodfit to the data (the linesaccountfor 98% of the variancein thresholds).An F testrejectedthe null hypothesisthat therewereno ageeffects(i.e., that the slopesof the lines in Fig. 2 were equal to zero) for all of the data [F(5,20) = 21.23,p