Pure-tone masking has trained the eye to recognize that large interaural differences signal the need for masking. You’ll need to learn new strategies in order to not miss the need for speech masking. You will need to think about the presentation level – and compare that to the non-test ear’s best bone-conduction thresholds. Because you are testing at a level that is well above threshold, and well above the non-test ear bone-conduction thresholds, you will frequently need to mask.

It’s easier (requires less hearing) to repeat spondee words from a list of a handful of choices than it is to correctly recognize what word was presented during word recognition testing. So for word recognition testing, a little bit of crossover might not truly improve the word recognition score. However, audiologists tend to be cautious and if there is any possibility of the non-test ear participating then contralateral masking is used. This means sometimes we will mask when it wasn’t essential. Figure 10-3 offers an example where the minimal cross hearing during word recognition testing does not do much to aid in figuring out which word was heard. Very little speech energy would be heard in the left ear. If this patient scored even modestly well (e.g. perhaps 40% or better), without contralateral masking having been used, then I would doubt that the non-test ear hearing (alone) could explain that score. Consider the degree to which the speech could be suprathreshold in the non-test ear. If it is marginal, as in Figure 10-3, then it is not an aid to word recognition. However, it doesn’t hurt to mask to eliminate any possibility that the non-test ear is contributing to the word recognition.

Figure 10-3. There is no need to mask for right ear spondee testing. The expected spondee threshold is about 60 dB HL. Crossover would be at 0 dB HL and inaudible. However, if testing word recognition at 85 dB HL in the right ear, the crossover (25 dB HL) is audible. Masking should be conducted, but a very low word recognition score would be expected if only the cross-hearing is contributing to word understanding. Only the speech energy below ~600 Hz is audible in the non-test (left) ear cochlea.

Audiologists may test word recognition at high intensity levels in order to determine if word recognition performance is poor or has decreased when intensity was increased – a sign of potential retrocochlear involvement. Since high test stimulus levels are used, crossover is high; participation of the contralateral ear would invalidate test results. Figure 10-4 illustrates.

Figure 10-4. Crossover of the right ear 100 dB HL speech signal to the non-test (left) ear could significantly improve the word recognition score. Even though the asymmetry is not large, masking when testing word recognition at high intensities is necessary.

Even with symmetrical hearing, high intensity speech testing requires masking. Even if the right ear in the audiogram in Figure 10-4 had the same hearing as the left ear, if word recognition were assessed at 100 dB HL, there will be the same crossover to the non-test ear

• To be efficient with spondee threshold formula masking, you want to make sure that you are masking when needed “from the start.” If you obtain the unmasked spondee threshold, then recognize the need for masking, you have to do extra testing. To avoid this possibility, consider the highest spondee threshold you would expect. The two-frequency average plus about 10 dB would be a good estimate. You want to go 10 dB “above” your estimate so that, if the spondee threshold is a bit higher than your estimate, you don’t have to go back and mask.
• Subtract 60 dB, the interaural attenuation value (for insert earphones, 50 for TDH).
• Compare this level to the non-test ear bone-conduction thresholds at 500 to 8k Hz (include 250 Hz if obtaining a speech detection threshold). If the probable bone-conduction thresholds are lower – you need to mask.
• If immittance is normal, use the NTE air-conduction thresholds as your guide for whether masking is needed. To be safest, consider the “what if” – what if the bone-conduction threshold is 10 dB lower? Use that as your estimate, again, so that you don’t obtain the unmasked threshold, then after obtaining bone-conduction thresholds, realize that you did have to mask. If immittance is abnormal and/or symptoms suggest conductive involvement, then most likely you would assume 0 dB HL bone-conduction thresholds, or perhaps even lower if the patient is young.

As with formula masking for pure tones, guessing the threshold will help you be efficient during testing. The pure-tone average should help you predict the spondee threshold you will eventually obtain. To reiterate, you won’t want to use exactly the predicted threshold in your minimum masking formula. If threshold comes in even a little higher, that would mean you have undermasked, and would need to increase the masking level. It is better to build the “what if the spondee threshold is a bit higher than estimated” into the formula.

Masking noise presented to the non-test ear is attenuated by any conductive loss in that non-test ear, so if you have any reason to believe there is conductive loss, estimate it (and use the worst case scenario – that it is as large as it could realistically be.) Add that into the minimum masking formula. As for pure-tone testing, if the air-bone gap is the 5 dB or so air-bone gap that is due to test-retest variability, those don’t need to be accounted for. When immittance is abnormal, then you should be very concerned, and the safest course is to assume bone-conduction thresholds are normal and calculate the air-bone gap. (As with pure-tone masking, you might want to switch to a MMax approach.)

How much masking is absolutely safe? If none of the crossback is audible, that is ideal. Assume a 60 dB IA at each frequency.

We can safely present up to 55 dB above the best bone-conduction threshold. If we go all the way up to 60 we risk some crossback interference.

So, the formula is MMax = Best TE bone-conduction threshold (500 to 8 kHz) + 55 dB

This formula holds the same for both spondee threshold and for word recognition testing, but as an upcoming section will discuss, a little bit of cross back may not hurt. As long as the speech signal is well above the crossback, the speech is still audible.

When thinking about the potential crossback and the number to use as “Best TE bone threshold”, think about cochlear sensitivity rather than the maximum output of the audiometer. For example, in Figure 10-5 below, the cochlear sensitivity at 500 Hz is probably 80 dB HL, which is higher than the audiometer bone-conduction circuit can produce. (Overmasking is dictated by the noise interfering with the test ear’s hearing, so that is what we need to consider. You won’t have overmasking unless the crossback is 80 dB or higher.) It would be a good idea to consider the possibility of a slight conductive loss, however. If you wish to be “extra safe” you could use 10 dB below the guessed best bone-conduction threshold.

The “Down 20 rule” works most times – the common exception is NTE significant conductive loss, where undermasking can occur. TE conductive loss can cause the “Down 20” rule to have problems, but you would need a near true maximum conductive loss before overmasking becomes a concern. The “Down 20” formula is simple (and therefore less prone to error) so it is recommended for the “plain vanilla” cases of asymmetrical sensorineural hearing loss, which is what you see most often in most clinical situations.

To review, if you have NTE conductive loss, then “Down 20” can undermask. Base your calculations on Minimum Masking Levels: go above that.

If the test ear has conductive loss, double check. Calculate MMax and make sure that it is below your “Down 20” level is below MMax. The test ear conductive loss has to be very large before overmasking becomes a concern when obtaining a spondee threshold.

If you have significant bilateral conductive loss, you will need to carefully calculate minimum and maximum. Similar to pure-tone testing with bilateral conductive loss, a masking dilemma may occur when trying to obtain the spondee threshold.

Although you need to mask for word recognition testing more often than for spondee threshold testing, at least you know the presentation level since you choose it! That eliminates that ambiguity you had with spondee threshold testing (what if it comes in a bit worse than I guessed?) To determine the need for masking, take the level of speech you plan to use for word recognition testing, subtract 60 dB. If there is any reason to believe that the non-test ear bone-conduction thresholds (in the 500 to 8k Hz range) are lower than that, then mask.

The minimum masking formula changes from what is used for spondee threshold testing. With spondee testing, you entertained the possibility of the spondee threshold coming in a bit higher than your best guess, and adjusted the masking level to account for that. You don’t do that for word recognition testing, since you know exactly the level you will use when testing, so the formula is presentation level – 60 dB IA + 10 dB pad + largest significant NTE air-bone gap (500 to 8k Hz) which simplifies to presentation level – 50 dB + largest significant NTE air-bone gap (500 to 8k Hz).

The maximum masking formula is the same as for spondee threshold testing: Find the best estimated bone-conduction threshold (and guess low rather than high to be safe, you don’t want to overmask). Add 55 dB. While this is the “ideal” maximum, a little bit of crossback isn’t going to lower the test ear word recognition score. If you can avoid going above MMax, please do so, but if the crossback is well below the perceived loudness of the speech signal, then it will not lower the word recognition score. We will examine this idea further below.

The simple “Down 20” formula remains simple: Subtract 20 from the presentation level. As with spondee threshold testing, this formula risks undermasking with NTE conductive loss, and overmasking with TE conductive loss, but as we will explore further below, the overmasking probably won’t interfere with understanding the monosyllabic words presented to the test ear.

As the examples above have introduced, crossback isn’t always “the end of the world.” This concept applies for spondee threshold testing as well, but word recognition testing involves presenting supra-threshold level stimuli, requiring higher masking levels, and greater crossback levels are seen.

If some of the noise crosses BACK to the test ear cochlea, that can interfere with word recognition, but it does not necessarily do so. Examine Figure 10-15.

---

Figure 10-15. The right ear has a large conductive overlay. Word recognition testing is conducted at 110 dB HL. The crossover can be heard in the left ear. The minimum masking level = 110-50: 60 dB EM. The maximum masking level = 15+55 = 70. Using anything within that narrow range of 60 to 70 dB EM would be fine, but let’s assume the audiologist was having an off day and used the “Down 20” formula, presenting 90 dB of noise to the left ear. As shown on the bottom of the figure, this could cause some crossback to the cochlea, which is audible at and below 750 Hz. Would it degrade word recognition performance? No, that’s unlikely. The sensation level of the speech signal (red arrows on the top figure) is above the sensation level of the crossback (blue arrows on the bottom of the figure.)

It is possible to have a masking dilemma for word recognition testing – the same situation as creates a masking dilemma for pure-tone testing: bilateral maximum or near maximum conductive loss. Refer to Figure 10-16.

Figure 10-16. Immittance test results are consistent with bilateral conductive loss: pure-tone testing with masking presents with a masking dilemma. So will spondee threshold and word recognition testing. Assume that you test left ear word recognition at 100 dB HL, which will be about 30 dB sensation level. The minimum masking level = 100-50+largest NTE ABG (about 70 dB) = 120. The crossback level (120-60) = 60. Since the loss is conductive, that means that the crossback is at about 60 dB sensation level. The noise that crosses back to the test ear cochlea will be louder than the speech sensation level (30 dB SL). The person won’t hear the words, let alone be able to repeat them. You’ll have the same problem with finding the spondee threshold. If the left ear spondee comes in at 65 dB HL, our formula recommends to use (65-40+~70) at minimum 95 dB EM, which could crossback at 35 dB EM, which will prevent hearing. (Remember, the cochlea does the speech recognition – the level of speech reaching the cochlea after attenuation by the conductive loss is about 0 dB HL. The crossback of 35 is more than enough to prevent you from measuring a 70 dB spondee threshold.) What would happen? You would find an spondee threshold perhaps of 40 dB HL instead – elevated by the crossback. Then you would recognize that you aren’t using enough masking noise since the spondee threshold was not what was expected. You’d increase the noise – and that would cause more overmasking, elevating the spondee threshold further.

When you have a potential masking dilemma, my recommendation is to calculate your minimum and maximum levels. If your minimum is above the maximum, think about the amount of crossback that could be created when you use that minimum level – and think about it frequency by frequency. Compare the crossback to the sensation level of the stimulus. If your sensation level is higher than the crossback by 10 dB or more, then the crossback likely has little or no effect

If there is more crossback, e.g. crossback is at 30 dB above the bone-conduction thresholds of the test ear, and the word recognition signal is 35 dB above the air-conduction thresholds, then the word recognition score is probably lowered. This is particularly true for those with sensorineural loss – their word recognition scores are lowered in the presence of competing signals. If you cannot lower the noise without undermasking, then I suggest that you make a note in your report, e.g. “Contralateral masking noise crossback potentially lowered the word recognition score.” You could also test unmasked and note “Because of the masking difficulties/dilemma potentially reducing the word recognition score, unmasked testing was conducted, but is influenced by the non-test ear participation. True word recognition performance is likely in the range between the masked and unmasked scores.”

When masking for speech, particularly word recognition testing, you must be vigilant about the need to mask. Remember to determine if the signal level – 60 dB is above the NTE best bone-conduction threshold (in the range of 500 to 8000 Hz). If so, masking is needed.

Stay away from the simple “stimulus level – 20” rule if the non-test ear is conductive. It works wonderfully for sensorineural losses, and is simple to use. When the test ear has conductive loss, consider whether the crossback might interfere. (Noise level – 60: is that significantly above the TE bone-conduction thresholds? If so, determine the stimulus sensation level. You are still OK if the stimulus is at least 10 dB above the crossback level.) Be very careful with significant NTE conductive loss, there it is better to use the minimum masking formula. The Down 20 rule tends to cause undermasking with NTE conductive components

We want to avoid having to raise the masking noise if the spondee threshold is slightly higher than predicted. Take your best estimate of the spondee threshold and adjust the formula by 10 dB more, in case the spondee threshold is a bit higher than your guess, which gives us the formula Minimum Level = Expected Spondee Threshold – 40 + Largest Significant NTE Air-Bone Gap (500 to 8k Hz).

For word recognition testing, you know precisely what the stimulus intensity is, so the formula is Minimum Level = Expected Spondee Threshold – 50 + Largest Significant NTE Air-Bone Gap (500 to 8k Hz).

Since the minimum speech IA is 60 dB, the maximum masking level formula is Max Mask = Best Test Ear Bone-Conduction Threshold (500 to 8k Hz) + 55 dB. If you want to be conservative, you can lower your estimate of the bone-conduction threshold by 10 dB (in case there are minor air-bone gaps).

Remember that some crossback and overmasking may not be a problem, especially with sloping losses. If the crossback is only audible in the lowest frequencies, then even for spondee threshold testing, the results may not be invalidated. With word recognition testing, you are presenting at a level that is suprathreshold. If the crossback is well below the level of the speech at the test ear cochlea, then the crossback doesn’t affect the word recognition score much if at all. Ideally, you would examine the sensation level of the noise above the test ear bone-conduction thresholds. Compare to the sensation level of the speech above the test ear air-conduction thresholds. You would want the speech sensation level to be at a minimum 10 dB louder. If undermasking problems prevent you from lowering the speech level when this criterion is not met, document that the word recognition scores may be lowered due to overmasking.