masking redux i: an optimized masking method · t he topic of masking hardly generates the same...

J Am Acad Audiol 15:17–28 (2004)

17

AbstractThis is the first in a series of two papers on masking. The objective of thesepapers is to develop a masking protocol that provides valid measures of thresh-old and is, in general, faster than the plateau method. In this paper, a maskingmethod is developed that can replace the traditional plateau method in mostmasking situations. This new method is optimized to require fewer maskinglevels than the plateau method. As a result, it can be significantly faster thanthe plateau method. There are some masking situations in which this opti-mized method cannot be used. This limitation will be evaluated in the secondpaper and an alternate masking strategy provided for use when the optimizedmethod is not appropriate.

Key Words: Audiometry, auditory thresholds, masking, optimized method,plateau method

Abbreviations: A1 = initial masking level (air conduction) for the optimizedmethod; ABG = air-bone gap; ABGN = actual air-bone gap in the nontest ear;AC = air conduction; ATN = air-conduction threshold in the nontest ear; ATT= air-conduction threshold in the test ear; B1 = initial masking level (bone con-duction) for the optimized method; BC = bone conduction; BTT =bone-conduction threshold of the test ear; IA = actual interaural attenuation;IM = initial masking level for the plateau method; MEM = minimum effectivemasking; MIA = assumed minimum interaural attenuation; NTE = nontest ear;TE = test ear; XUM = maximum usable masking

SumarioEste es el primero de una serie de dos trabajos relacionados con elenmascaramiento. El objetivo de estos trabajos es desarrollar un protocolode enmascaramiento que proporcione mediciones válidas del umbral y quesea, en general, más rápido que el método del plateau. En este trabajo, sedesarrolla un método de enmascaramiento que reemplaza el tradicionalmétodo del plateau, en la mayoría de las situaciones que requierenenmascaramiento. El nuevo método está optimizado para requerir menosniveles de enmascaramiento que el método del plateau. Como resultado, puedeser significativamente más rápido. Existen algunas situaciones donde estemétodo optimizado no puede ser usado. Esta limitación será evaluada en elsegundo trabajo, donde se aportará una estrategia alternativa deenmascaramiento para usar cuando el método optimizado no sea apropiado.

Palabras Clave: Audiometría, umbrales auditivos, enmascaramiento, métodooptimizado, método del plateau o meseta

Abreviaturas: A1 = nivel de enmascaramiento inicial (conducción aérea)para el método optimizado; ABG = brecha aéreo-ósea; ABGN = brecha aéreo-ósea real en el oído no evaluado; AC = conducción aérea; ATN = umbral de

Masking Redux I: An Optimized MaskingMethod

Robert G. Turner*

*Louisiana State University Health Sciences Center

Reprint requests: Robert G. Turner, Department of Communication Disorders, Louisiana State University Health Sciences Center,1900 Gravier St., Box G6-2, New Orleans, LA 70112; Phone: 504-568-4339; Fax: 504-568-4352; E-mail: [email protected]

The material in this paper has been presented at the American Academy of Audiology 15th Annual Convention, Philadelphia, PA,April 2002 and the American Speech-Language-Hearing Association Convention, New Orleans, LA, November 2001.

90305 Cassidy text 2/12/04 10:22 AM 7

The topic of masking hardly generatesthe same excitement amongaudiologists as hair cell regeneration,

digital hearing aids, or the Au.D. Studentsrank masking with impedance and the decibelas the three most tedious, confusing subjectsin their graduate education. Whataudiological topic could be more boring thanmasking? However, audiologists still spendmuch of their time generating audiograms,and masking is sometimes required to obtainaccurate measures of threshold. Survey datafrom Martin et al (1994) suggest that all isnot well in the world of masking. In fact,they state, “despite the large body of researchregarding clinical masking, many audiologistsuse improper determinations for the need tomask and/or fail to mask using a logicalmethod” (Martin et al, 1994, p. 26). A morerecent survey (Martin et al, 1998) alsodemonstrates that many audiologistscontinue to use inappropriate maskingprocedures.

The most common masking procedure is the plateau method; 58% of surveyrespondents report using some variant of thistechnique (Martin et al, 1994). Disturbingly,more than 40% did not use the plateaumethod. Seven percent used an “arbitrary”level for masking, and 26% categorized theirprocedure as “other.” The plateau method iswell established in the research literatureand works quite well when properlyadministered. A review of audiology textbooksindicates that the plateau method is theprocedure consistently described.

Why is it that so many audiologists do notuse the plateau method? The plateau methodhas many advantages, but it has onesignificant disadvantage. It can be extremelytime consuming. Our experiences indicatethat students who are taught the plateaumethod frequently abandon this techniquewhen they leave the unique environment ofthe academic clinic. In many clinical settings,time for testing is significantly limited. Theplateau method appears impractical in these

situations, and students, or graduates, adopt“shortcut” masking procedures, typicallyacquired from supervisors or coworkers. Theproblem is that these shortcut proceduresmay not be based on theoretical principles orvalidated in any way. These procedures maywork in certain situations, suggesting theirvalidity, but fail in other situations with theaudiologist unaware of these failures.

The objective of this paper is to describea masking method that (1) yields validmeasures of threshold, (2) is as simple toperform as the plateau method, and (3) ismore time efficient than the plateau method.This is the first step in developing arecommended masking protocol that can beused in all masking situations. The maskingmethod developed in this paper can replacethe plateau method in most, but notnecessarily all, masking situations. Thelimitations of this new method will beevaluated in the second paper in this series(Turner, 2003).

This paper is not an introductory tutorialon masking; it is assumed that the reader hasa basic understanding. It will be necessary,however, to review some principles, as theyare needed to describe the derivation of thenew method.

THE PLATEAU METHOD

The plateau method was first describedby Hood (1960) over four decades ago.

Interestingly, Hood presented this techniquefor bone-conduction testing and never usedthe word “plateau” in his article. Hood’sprocedure was adapted to air-conductiontesting and became known as the “plateaumethod.” The plateau method is typicallydescribed using a masking diagram as shownin Figure 1a. A tone is presented to the testear (TE) by air conduction (AC) or boneconduction (BC). A masking noise is presentedto the nontest ear (NTE) by air conduction.The noise is calibrated in terms of effectivemasking. For this paper, the definition of

Journal of the American Academy of Audiology/Volume 15, Number 1, 2004

18

conducción aérea en el oído no evaluado; ATT = umbral de conducción aéreaen el oído evaluado; B1 = nivel de enmascaramiento inicial (conducción ósea)para el método optimizado; BC = conducción ósea; BTT = umbral de con-ducción ósea en el oído evaluado; IA = atenuación interaural real; IM = nivelinicial de enmascaramiento para el método del plateau; MEM = enmas-caramiento efectivo mínimo; MIA = supuesta atenuación interaural mínima;NTE = oído no evaluado; TE = oído evaluado; XUM = enmascaramientomáximo utilizable

90305 Cassidy text 2/12/04 10:22 AM Page 18

Masking Redux I/Turner

19

“effective masking” is consistent with ANSIS3.6, 1989. For example, a 60 dB HL tonewould just be audible with an effective maskerof 60 dB HL presented to the same ear. Thisdiffers slightly from another definition

occasionally used in which a 60 dB HLeffective masker would just mask a 60 dB HLtone (Martin, 1986).

The traditional masking diagram depictsthe threshold responses of the individual asthe masking level is increased (Figure 1a).(An explanation of the audiometric symbolsused in this and other figures is given inFigure 2.) In this example, the actual ACthreshold of the TE is 90 dB HL, and theactual BC threshold is 50 dB HL. The ACresponse at 60 dB HL occurs because ofcrossover to the NTE. The actual AC and BCthresholds of the NTE are 10 dB HL.

The masking curve can have severaldistinct regions defined by inflection pointsin the curve. Masking diagrams are easilyconstructed once the locations of the inflectionpoints are determined. The first inflectionpoint occurs when the masking level equals

Figure 1. Traditionalmasking diagram. (a)Modified masking diagramand frequency diagram. (b)The modified maskingdiagram inverts the ordinate(Hearing Threshold) so asto be consistent with theaudiogram. Both maskingdiagrams show thresholdsas a function of effectivemasking level. The shadedarrows represent maskinglevels for the plateaumethod. The frequencydiagram represents onefrequency, the test frequency,on the audiogram. SeeFigure 2 for a key to theaudiometric symbols. IM:initial masking level.

Figure 2. Key to the audiometric symbols used inthe figures.



20

the AC threshold in the NTE. If the maskeris presented at a level below 10 dB HL, thenthe masker would not mask the NTE. Thus,this would be a region of “no masking.” Ofcourse, this is not common clinical practice.When the masking level exceeds the ACthreshold of the NTE, there would be maskingof the NTE. If the unmasked threshold wasdue to crossover to the NTE, then thethreshold would increase when the NTE ismasked. This would be the region called“undermasking.” In this region, the thresholdof the NTE is being elevated by the masker,but the tone has not reached the threshold ofthe TE. Thus, the tone is audible in the NTEbut is not audible in the TE.

The next region is the “plateau.” Theplateau is the masking range between the twoinflection points corresponding to theminimum effective masking (MEM) and themaximum usable masking (XUM). Minimumeffective masking is the minimum maskinglevel that makes the test tone at the thresholdof the TE inaudible to the NTE, and it occursat the first inflection point. Maximum useablemasking is the maximum masking levelbefore the masker crosses over and elevatesthe threshold in the TE. In the plateau, thetone is audible to the TE. The tone is notaudible in the NTE because of the masker,and the masker is not sufficiently intense tocross over and elevate thresholds in the TE.The threshold measured in the plateau isthe actual threshold of the TE. The maskinglevels at which the minimum effectivemasking and maximum usable masking occurcan be calculated using equations firstprovided by Liden et al (1959). Theseequations are shown below with somemodification. For AC testing

MEM(A) = ATT + ABGN - IA (1)XUM(A) = BTT + IA (2)

where ATT is the actual AC threshold of theTE, ABGN is the actual air-bone gap (ABG)in the NTE, BTT is the actual BC thresholdin the TE, and IA equals the actual interauralattenuation for air conduction. For BC testing,

MEM(B) = BTT + ABGN (3)XUM(B) = BTT + IA (4)

The original equations of Liden et al ignoredthe occlusion effect that will be considered inmore detail in the second paper.

As the masking level is increased, thethreshold tone remains audible in the TEuntil the masker crosses over to the TE at asufficient level to mask the response of theTE. As the masking level continues toincrease, the masked threshold of the TEcontinues to increase. In this region, called“overmasking,” the tone is not audible in theNTE but is audible in the TE at an elevatedlevel above the actual threshold.

The basic procedure as described byHood:

1. Establish unmasked thresholds.2. If required, apply masking to the

NTE and reestablish threshold.3. Increase the masking level in 10 dB

steps until the measured threshold“remains constant with further additional incremental steps of 10 dB” (Hood, 1960, p. 1227).

Several variations on the plateau methodhave been suggested. It is worthwhile toreview some of these modifications. The firstissue is when to mask. For AC thresholds,Hood recommended masking when thedifference in AC thresholds exceeds 50 dB.The accepted current recommendation is tomask when the AC threshold in the TEexceeds the BC threshold of the NTE by aspecified amount, the assumed minimuminteraural attenuation (MIA). For supra-aural headphones, this value is typically 40dB, independent of frequency. This value islarger for insert phones. (Yacullo, 1996, 21,128). In this paper, we will assume thatsupra-aural headphones are being used. Thisis the more difficult masking situation thanwith insert phones. Any procedure that willwork with supra-aural headphones will workwith insert phones.

For BC testing, the interauralattenuation can be as small as 0 dB. Thus,some, including Hood, recommend alwaysmasking BC thresholds. There is no problemwith this strategy except that it takes moretime. Another recommendation, the one usedin this paper, is to mask when the ACthreshold in the TE exceeds the BC thresholdin the NTE by more than 10 dB (Yacullo,1996, 26). If available, the masked BCthreshold in the NTE would be used;otherwise, use the unmasked threshold. It isnot that small ABGs are always unimportant.The rationale is that the BC threshold canshift up to 10 dB due to test variability and



21

central masking. Thus, the closure of anapparent 10 dB ABG would not necessarilyprove that there is no ABG.

Hood did not specify the initial maskinglevel (IM). Martin (1974) has recommended

IM = ATN + 10 dB (5)

where ATN is the AC threshold of the NTE.The 10 dB is a safety factor to account for testand subject variability and to insure thatthe masker actually masks the NTE. Anadditional factor for the occlusion effect maybe required for BC testing.

One variation that can reduce the timerequired to perform the plateau method is tonot reestablish threshold using the traditionalup-down search for threshold (Yacullo, 1996,73). Instead, the tone is presented once. Ifthere is a response, that is taken as thresholdat that masking level, and the masking levelis increased. If there is no response, then thetone is increased in 5 or 10 dB steps andpresented once at each level until there is aresponse. This response indicates thethreshold at that masking level. If this“single-tone” procedure is used to reestablishthreshold and identify the plateau, then thethreshold in the plateau, which is the actualthreshold of the ear, should be measuredusing the more accurate up-down procedure.

Hood recommended increasing themasking level in 10 dB steps, but some(Roeser and Clark, 2000) recommend 5 dBsteps. The use of 5 dB steps would essentiallydouble the number of masking levels andincrease testing time. Using 5 dB steps wouldbe useful if the plateau were narrow anddifficult to define.

If the unmasked threshold does notchange with the initial masking, somerecommend accepting the unmaskedthreshold as the actual threshold. This wouldreduce testing time. It is generallyrecommended, however, that the maskinglevel be increased at least once to ensurethat the NTE is adequately masked(Studebaker, 1967). Hood does not specifythe number of increases of masking levelneeded to define the plateau region. It isgenerally recommended that the maskinglevel be increased over a range of at least 15to 20 dB with little change in tone threshold(Yacullo, 1996, 73).

The time required to perform the plateaumethod is an issue in this paper. Obviously,

that time is a function of the specific plateauprocedure used. Reestablishing thresholdusing the traditional up-down search andusing 5 dB masker steps would produce theslowest procedure. Using a single tonepresentation at each tone level to reestablishthreshold, as discussed above, and using 10dB masker steps would produce the quickestprocedure. The masking method described inthis paper will be compared to the plateaumethod. For this purpose, the plateau methodwill be defined as the following:

1. Set the initial masking level (IM) 10dB above the AC threshold of the NTE and reestablish threshold.

2. The masking level is increased 10 dBand threshold reestablished.

3. When the masking level is increasedtwice (20 dB range) with no change(poorer) in threshold, the masker isin the plateau, and the measured threshold is the actual threshold.

The plateau method is illustrated by theshaded arrows in the masking diagram(Figure 1a). Each arrow represents a maskerpresentation, and the arrows conform to thedefinition above.

MASKING AND FREQUENCYDIAGRAMS

Masking diagrams will be use extensivelyin this and the subsequent paper, but

with a modification as shown in Figure 1b.The ordinate (Hearing Threshold) is reversedfrom that in Figure 1a. This was done so asto be consistent with the audiogram; that is,threshold increases in the downwarddirection. The plateau method is alsorepresented with shaded arrows in themasking diagram in Figure 1b. In all maskingdiagrams, the right ear will be the TE and theleft the NTE.

To the left of the masking diagram is afrequency diagram, corresponding to onefrequency on an audiogram. The frequencydiagram will be used to represent a varietyof information. The various masking levelsused are represented by the arrows to theright of the axis. These correspond to thearrows on the masking diagram. UnmaskedAC thresholds are shown using conventionalsymbols (right—60 dB HL; left—10 dB HL).The unspecified symbol for unmasked bone(10 dB HL) is used because an unmasked



22

BC threshold cannot, in general, be associatedwith a particular ear. Masked thresholds (70,80 dB HL) are shown using conventionalsymbols and actual thresholds (AC—90dBHL; BC—50 dB HL) are represented by filledsymbols. If the actual threshold equals theunmasked threshold, then only the unmaskedthreshold is shown so as to reduce clutter. Itis also possible to indicate what maskinglevel(s) produces a particular maskedthreshold. In Figure 1b, IM resulted in amasked AC threshold in the right ear of 70dB HL. Masking levels 3, 4, and 5 elevatethreshold to the actual threshold, becausethey are in the plateau region.

The masking and frequency diagrams aregenerated based on theoretical considerations.Two factors that could influence actualthreshold measurements are ignored. Theseare test variability and central masking. Themeasurement of threshold can vary, typicallyabout ± 5 dB. In addition, central masking canincrease threshold by an average of 5 dB. Theimplications of these factors to maskingprocedures are discussed later in this paper.They would, however, add unnecessarycomplexity to the masking and frequencydiagrams and have, therefore, been ignored.

OPTIMIZED MASKING METHOD

The new masking method is similar to theplateau method and should be at least as

easy to perform. It is called the optimizedmethod because it has been optimized inseveral ways to reduce testing time. In manymasking situations, this optimized methodcan significantly reduce the number ofmasking levels required to reach the plateauand determine threshold. Fewer maskinglevels result in less testing time. This methodis described below and is applicable for ACand BC testing.

1. Measure an unmasked AC thresholdfor each ear and an unmasked BCthreshold.

The unmasked BC threshold couldcorrespond to the actual BC threshold of theTE, the NTE, or both.

2. At each frequency, determine ifmasking is required using conventionalcriteria.

Because there is a single unmasked BCthreshold, it is assumed, until shownotherwise, that this represents the actualBC thresholds of both ears. To determine theneed for masking AC or BC thresholds, theunmasked AC threshold is compared to theunspecified unmasked BC threshold. Whenthis difference equals or exceeds the MIA,then masking is required for the ACthreshold. When this difference exceeds 10dB, then masking is required for the BCthreshold.

3. Set the initial masking level (A1 orB1) equal to the AC threshold of theTE minus 10 dB and reestablishthreshold.

A1 is the initial masking level for air-conduction testing. B1 is the initial maskinglevel for bone-conduction testing. These twoinitial levels will always be equal. Subsequentmasking levels may differ for air- and bone-conduction testing. In general, the level of B1is sufficient to compensate for the occlusioneffect when testing low frequencies, and noadditional correction is needed. This isdiscussed in detail in the second paper.

With this strategy, the initial masker is setrelative the threshold of the TE. This is amajor difference from the plateau methodwhere the initial masking level (IM) typicallyequals the AC threshold of the NTE plus 10dB. One way the new method is optimized isthat it attempts to use the maximum possiblemasking without overmasking. To accomplishthis, the masking level must be referenced tothe threshold of the TE, not the NTE. Theplateau and other masking methods referencethe initial masker to the threshold of the NTE.

The optimized method is based on theprinciple that setting a masking level 10 dBbelow the AC threshold of the TE will neverresult in overmasking. This is illustrated inFigure 3a. In this example, we have used anMIA equal to 40 dB. The unmasked ACthreshold of the TE (right) is at 70 dB HL, 60dB above the unmasked BC threshold. Sincethe MIA is 40 dB, masking is needed. In thismethod, the initial masking for air conduction(A1) and bone conduction (B1) is set at 60 dBHL, which is 10 dB below the AC thresholdof the TE. Thus, the masking level is 50 dBabove the unmasked BC threshold, and thisexceeds the MIA. Can this level of maskerproduce overmasking? Even though we used



23

an MIA of 40 dB to determine the need formasking, the IA in this example must be atleast 60 dB. An explanation of why this istrue will be provided later. Overmaskingwould occur if the masker at 60 dB HL couldcross over to the TE at a level sufficient tomask the TE, that is, at a level above theactual BC threshold of the TE. The worstcase for overmasking would occur if the actualBC threshold of the TE equaled the unmaskedBC threshold. Even in this situation, however,overmasking would not occur. The masker at60 dB HL would be reduced by the IA that isat least 60 dB. Thus, the masker would reachthe TE at a maximum level of 0 dB HL, lessthan the best possible BC threshold of the TE(10 dB HL). This argument applies equallywell to BC testing. Since overmasking occursby the same mechanism in AC and BC testing,a masking level that does not overmask for ACtesting will not overmask for BC testing.

In the example in Figure 3a, there is anapparent ABG of 60 dB. This tells us that theIA must be at least 60 dB. To demonstratewhy this is true, we must consider twopossibilities: (1) the actual BC threshold of theNTE equals the unmasked BC threshold; (2)the actual BC threshold of the TE equals the

unmasked BC threshold. Consider the firstpossibility that the unmasked BC thresholdcorresponds to the NTE. In this situation,there is a 60 dB difference between theunmasked AC threshold of the TE and the BCthreshold of the NTE; thus the IA must be atleast 60 dB. If the IA were less than 60 dB,the unmasked AC threshold of the TE wouldhave occurred at a better threshold, forexample, 50 dB HL.

Next, consider the second possibility thatthe unmasked BC threshold equals the actualBC threshold of the TE indicating an actualABG of 60 dB. Assume that the actual BCthreshold of the NTE is poorer, for example,30 dB HL. Could the unmasked AC thresholdin the TE (70 dB HL) be due to crossover tothe BC threshold of the NTE (30 dB HL)indicating an IA of 40 dB? If the IA were 40dB, then an AC signal to the TE at 70 dB HLwould stimulate the cochlea of the NTE bybone conduction at a level of 30 dB HL. Thisis at the BC threshold of the NTE and wouldjust be audible. What is often not recognizedis that this signal would also reach the cochleaof the TE at about 30 dB HL because theinteraural attenuation for bone conduction isas little as 0 dB and the NTE is being

Figure 3. Justification of strategies used in optimized masking protocol. See text for details. See Figure 2 fora key to the audiometric symbols. A1: initial masking level for air-conduction testing using the optimizedmethod; A2: additional masking level for air-conduction testing using the optimized method. B1: initial masking levelfor bone-conduction testing using the optimized method; B2: additional masking level for air-conduction testingusing the optimized method.


stimulated by bone conduction. Stateddifferently, a signal that can sufficientlyvibrate the skull so as to stimulate the cochleaof the NTE at 30 dB HL will also stimulatethe cochlea of the TE at about 30 dB HL.Since 10 dB HL is the BC threshold of the TE,this signal would be 20 dB above that BCthreshold. Obviously, a lower-level signalwould be audible to the TE. If the IA were 40dB, then an AC signal to the TE at 50 dB HLwould stimulate the cochlea of the TE bybone conduction at a level of 10 dB HL andwould be audible. Thus, the unmasked ACthreshold of the TE would occur at 50 dB HL,not 70 dB HL, and the ABG would be 40 dB.Since the unmasked AC threshold of the TEis 70 dB with an ABG of 60 dB, the IA mustbe at least 60 dB. This illustrates afundamental principle upon which thismasking strategy is based. An ABG cannot begreater than the IA, although it can, of course,be less. The unmasked AC threshold in the TE(70 dB) cannot be due to crossover to the BCthreshold of the NTE (30 dB) because thatwould mean an ABG in the TE significantlygreater than the IA, which is not possible.

The apparent ABG in the TE in Figure 3ais 60 dB. For both possibilities, the IA had tobe at least 60 dB supporting the statementthat the IA cannot be less than the apparentABG. In general, the IA could be greater than60 dB because the unmasked AC threshold in

the TE could be the actual threshold and notdue to crossover to the NTE.

There is, of course, some test variabilityin the measurement of threshold and theeffectiveness of masking. That is why themasking level is set 10 dB below the ACthreshold of the TE. This provides a safetymargin to account for test and subjectvariability. Theoretically, the masking levelcould be set equal to the AC threshold of theTE and not result in overmasking.Referencing the masking level to thethreshold of the NTE can result in using lessthan the maximum permissible masking, asshown in Figure 4. Also, setting the maskinglevel some fixed amount, for example, 30 dB,above the threshold of the NTE could resultin overmasking as in the case of bilateralconductive loss.

This step calls for reestablishing thethreshold. As discussed above for the plateaumethod, threshold can be reestablished usingthe traditional up-down procedure or thesingle-tone procedure that can significantlydecrease testing time.

4. Determine the dB shift in thresholddue to the masker.

This is the difference between theunmasked threshold and the masked


24

Figure 4. Comparison ofplateau and optimizedmasking methods for deter-mining an AC threshold.The shaded arrows repre-sent masking levels for theplateau method. The solidarrows represent maskinglevels for the optimizedmethod. The plateaumethod requires sevenmasking levels to deter-mine threshold. The opti-mized method onlyrequires three maskinglevels. See Figure 2 for akey to the audiometricsymbols. A1: initial mask-ing level for air-conductiontesting using the optimizedmethod; A2, A3: additionalmasking levels for air-con-duction testing using theoptimized method. IM: ini-tial masking level for theplateau method.


threshold due to the initial masking level, A1or B1.

5. Increase the masking level anamount equal to the threshold shiftand reestablish threshold.

This new masking method is optimizedin a second way. It increases the masking levelthe maximum amount without overmasking.Increasing the masking level an amountequal to the threshold shift will not result inovermasking, as illustrated in Figure 3b. Inthis example, the threshold shifted 50 dB to120 dB HL in response to the masker; thus,the masking level is increased 50 dB to 110dB HL. Note that the masker is still 10 dBless than the masked AC threshold. Whentesting AC thresholds, this method keeps themasking level 10 dB below the masked ACthreshold in the TE. Because the thresholdin the TE shifted, we know that theunmasked threshold was due to crossover tothe NTE and the IA equals 60 dB. Since anABG cannot exceed the IA, the largestpossible ABG in this example is 60 dB. Thus,the best possible BC threshold in the TE is60 dB HL. The masker, which is at 110 dBHL, would reach the TE at a level of 50 dBHL, 10 dB below the BC threshold.Overmasking would not occur.

A similar argument can be made for BCtesting (Figure 3c). The initial masking level(B1) is 10 dB below the AC threshold of theTE and, thus, 50 dB above the unmaskedBC threshold. In this example, the BCthreshold in the TE elevates 50 dB inresponse to the masker. The masking level isincreased 50 dB to 110 dB HL. For BC testing,the masking level is always maintained thesame dB above the BC threshold, 50 dB inthis example. The IA is at least 60 dB; thus,the maximum crossover level of the maskerto the TE would be 50 dB HL, below the levelof the masked BC threshold. Overmaskingwould not occur.

Even if the unmasked threshold is theactual threshold and the shift is due to testvariability and/or central masking, increasingthe masking level an equal amount isappropriate. If the threshold shifts 5 dB dueto test variability, increasing the maskinglevel 5 dB should not overmask. Thesubsequent threshold should remain constantor improve, indicating actual threshold. Ifthe threshold shifts 5 dB due to central

masking, there is no way to determine this.Increasing the masking 5 dB will notovermask. Again, the subsequent thresholdwill be the appropriate threshold to record asthe actual threshold. There is disagreementas to whether there should be a correction forcentral masking. That issue is not addressedin this paper.

6. If threshold improves or does notshift, the masker is in the plateau, andthe actual threshold has beendetermined. If the threshold shifts,repeat steps 5–6.

If the threshold improves or does notshift with the initial masking level, then theunmasked threshold is the actual threshold.As noted above, it has been recommendedthat at least one additional masking levelbe used to identify the plateau. With thismethod, the initial masking level is usuallymuch greater than the AC threshold of theNTE, sufficient to shift threshold if theunmasked threshold is due to crossover to theNTE. However, it is generally acceptable touse a second level 5 dB to 10 dB greater toconfirm that the unmasked threshold is theactual threshold. The only limitation is thecase of significant bilateral conductive losswhere the plateau can be small. This situationis discussed in the next paper.

If the threshold shifts with the initialmasker level, then additional levels are useduntil the plateau is identified. If a singlestimulus presentation is used at each maskerlevel instead of reestablishing threshold, thenthreshold must be accurately measured oncethe plateau has been identified.

OPTIMIZED VERSUS PLATEAUMETHOD

Masking diagrams can be used toillustrate the optimized method and to

compare it to the plateau method. Theapplication to AC testing is shown in Figure4. From the frequency diagram, it is evidentthat there is a severe, unilateral,sensorineural loss in the right ear. The rightunmasked AC threshold is 45 dB greaterthan the unmasked BC threshold. We will usean MIA of 40 dB. Thus, we need to usemasking for the right AC threshold. Withthe plateau method, the IM would be 10 dBHL. The masking level would be increased in


25


10 dB steps until the plateau is defined. Thiswould require the use of seven masking levels,as shown in Figure 4 by the broad, shadedarrows.

With the optimized method, initialmasking level (A1) is set at 35 dB HL, 10 dBless than the AC threshold of the TE. Thismasking level is 35 dB above the AC thresholdof the NTE. Thus, we would expect a shift inmasked threshold up to 35 dB. In this case,the right AC threshold shifts 35 dB from 45dB HL to 80 dB HL. Because the thresholdhas shifted 35 dB, we increase the maskinglevel by 35 dB. This level (A2) produces a 10dB shift in threshold. Because the thresholdshift (10 dB) was much less than the increasein masking level (35 dB), we have strongevidence that A2 is in the plateau. The otherpossibility is that the masker passed over asmall plateau and is overmasking. That,however, would only occur with significantbilateral conductive loss.

We could take the threshold obtainedwith A2 as the actual threshold. Consistentwith the method, however, the masking levelis increased 10 dB (A3). The thresholdremains the same, confirming that themaskers A2 and A3 are in the plateau region.Even if the threshold shifts (poorer) 5 dBdue to test variability, we are still confidentthat we are in the plateau because themasking level has been increased 45 dB (A1

to A3), and the threshold has only shifted 15dB. If desired, the masking level could beincreased another 5 dB to confirm the plateau.This example demonstrates the advantage ofthe optimized method relative the plateaumethod. The optimized method requires threemasking levels to obtain actual threshold,whereas the plateau method requires seven.

The optimized method can also be usedfor BC testing (Figure 5). This is the samethreshold configuration as in Figure 4. Theunmasked thresholds indicate the possibilityof an ABG in the right ear greater than 10 dB.Thus, masking is required for the BCthreshold in the right ear. As it was for ACtesting, the initial masking (B1) is set at 35dB HL. This produces a threshold shift of 35dB; thus, the masking level is increased 35dB to 70 dB HL. This masker (B2) shifts theBC threshold beyond the limit of theaudiometer, which is assumed to be 60 dB HLfor this example. Using the optimized method,it took two masking levels to establish BCthreshold. Had the plateau method beenused, it would have required seven levels toshift the threshold beyond the limits of theaudiometer.

DISCUSSION

The objective is to recommend a maskingprotocol that is easy, fast, and valid. If


26

Figure 5. Comparison ofplateau and optimizedmasking methods fordetermining a BC threshold.The shaded arrows representmasking levels for the plateaumethod. The solid arrowsrepresent masker levels forthe optimized method. Theplateau method requiresseven masking levels todetermine threshold. Theoptimized method onlyrequires three maskinglevels. See Figure 2 for a keyto the audiometric symbols.B1: Initial masking level forbone-conduction testing usingthe optimized method; B2:additional masking level forbone-conduction testing usingthe optimized method. IM:initial masking level for theplateau method.


successful, then this protocol may be practicalin clinical environments where speed isessential. This protocol may be attractive toaudiologists who have adopted shortcutmasking procedures because the plateaumethod is perceived as being too slow. Thereis an additional advantage to reduced testingtime. Fewer masking levels at multiplefrequencies would reduce listener fatigue.This could be particularly useful with childrenand the elderly.

The first step is the development of anoptimized masking method that can replacethe plateau method in many maskingsituations. The procedure for the optimizedmasking method is as follows:

1. Measure an unmasked AC thresholdfor each ear and an unmasked BC threshold.

2. At each frequency, determine if masking is required using conventional criteria.

3. Set the initial masking level equalto the AC threshold of the TE minus10 dB and reestablish the threshold.

4. Determine the dB shift in thresholddue to the masker.

5. Increase the masking level an amount equal to the threshold shiftand reestablish threshold.

6. If threshold improves or does not shift, the masker is in the plateau,and the actual threshold has been determined. If the threshold shifts,repeat steps 5 and 6.

The optimized masking method is notas complex as it may appear. In procedure,it differs from the plateau method in only twosignificant ways. First, the initial maskinglevel is set 10 dB below the AC threshold ofthe TE. With the plateau method, initialmasker level is 10 dB above the AC thresholdof the NTE. Second, the masking level isincreased an amount equal to the maskedthreshold shift. With the plateau method,the masking level is increased a fixed amount,usually 5 or 10 dB. Both of these differencesare easy to remember; the optimized methodshould be as easy to administer as the plateaumethod.

The optimized method is faster than theplateau method because fewer masking levelsare required to reach plateau and determinethreshold. The actual time advantage,however, depends on threshold configurationand can be greater or less than that shown

in the examples above (Figures 4, 5). Thisissue will be examined in detail in the nextpaper.

As discussed above for the plateaumethod, threshold can be reestablished usingthe traditional up-down search or the single-tone procedure. There is no reason that thesingle-tone procedure cannot be employedwith the optimized method provided thresholdis reestablished with the more accurate up-down procedure once the plateau is identified.This strategy would increase the timeefficiency of the optimized method.

The optimized method is based onaccepted theoretical principles. One lesserknown principle is that an air-bone gapcannot be greater than the actual interauralattenuation that is determined by the type ofheadphone. Thus, a consequence of thisprinciple is that the headphone determinesthe maximum conductive loss, that is, ABG,that can be recorded on an audiogram.Historically, the maximum conductive losswas thought to be about 60 dB. This actuallyresulted from using supra-aural headphonesthat typically provided a maximum IA ofabout 60 dB. With the use of insert earphones,which can produce a greater IA, it is possibleto see ABGs in the 70 dB to 80 dB range.

Is the optimized method appropriate andvalid in all masking situations? In the nextpaper, the optimized method will be evaluatedfor a variety of threshold configurations.Masking diagrams will be used to determinethe ability of the optimized method to placethe masker within the plateau region and,thus, determine threshold. There are, in fact,some limitations on the use of the optimizedmethod. The exact nature of these limitationswill be described as well as an alternatemasking strategy that can be used in thosesituations where the optimized method isnot appropriate.

Acknowledgment. The author would like to thankDr. Robert Margolis and the two anonymousreviewers for their efforts and their valuable com-ments about this and the subsequent paper.

REFERENCES

American National Standards Institute. (1989).Specifications for Audiometers (ANSI S3.6-1989). NewYork: ANSI.

Hood J. (1960). The principles and practices of bone-conduction audiometry. Laryngoscope 70:1211–1228.


27


Liden G, Nilsson G, Anderson H. (1959). Masking inclinical audiometry. Acta Otolaryngol 50:125–136.

Martin F. (1974). Minimum effective masking levelsin threshold audiometry. J Speech Hear Disord39:280–285.

Martin F. (1986). Introduction to Audiology. 3rd ed.Needham Heights, MA: Prentice-Hall.

Martin F, Armstrong T, Champlin C. (1994). A surveyof audiological practices in the United States in 1992.Am J Audiol 3:20–26.

Martin F, Champlin C, Chambers J. (1998). Seventhsurvey of audiological practices in the United States.J Am Acad Audiol 9:95–104.

Roeser R, Clark J. (2000). Clinical masking. In: RoeserR, Valente M, Hosford-Dunn H, eds. AudiologyDiagnosis. New York: Thieme, 243–279.

Studebaker G. (1967). Clinical masking of the non-test ear. J Speech Hear Disord 32:360–371.

Turner R. (2003). Making redux II: a recommendedmasking protocol. J Am Acad Audiol 15:29–46.

Yacullo W. (1996). Clinical Masking Procedures.Needham Heights, MA: Allyn and Bacon.


28


masking redux i: an optimized masking method · t he topic of masking hardly generates the same...

Documents