Consonant confusion

For people with sloping audiograms (it may have been called ‘sensorineural’, ‘sensory-neural’, ‘nerve deafness’ etc.) what you hear can continue to sound as loud as ever but make less and less sense.

Each speech-sound (‘phoneme’) is made up of bursts of energy at a few frequencies (‘formants’). Simplifying a little, each phoneme is made up of a unique combination of 2 or 3 different frequencies. If you have a sloping hearing loss, you may continue to hear the lower formant(s) just as loud as you ever did, but not be hear the top formant(s). Unfortunately, two phonemes may be identical except for the top formant(s). So you are able to perceive the sound, you just can’t tell it apart from another phoneme with the same lower formant(s). For people whose sloping audiograms are progressive, the slope slides to the left on the audiogram over time. The more high frequency sound you lose, the more phonemes overlap.

Scientists researched this in the 1950s because it is important to the design of telephone systems. They researched speech discrimination (ability to tell phonemes apart) with ‘cutoff frequencies’ at different levels. The diagram below is based on those studies. The diagram shows several diagonal lines running from high at the left down to low at the right. You can compare those lines with your own sloping audiogram and choose the line that resembles your own audiogram slope. They only reported results for certain cutoff frequencies, so my apologies if your exact loss isn’t there. Also, the slope of the loss (22 dB per octave) is significant. A steeper or shallower slope, or a notch, is not the same. However, they didn’t report on those. It should at least be illustrative, and may give you an idea anyway; you’ll have to just try to get what you can out of it!

(The scientists did the same experiments with slopes going the other way, as though the person could hear high frequencies but not low ones. This does happen, but not as commonly, so I have not diagrammed it. It is worth noting that the nature of the confusion among consonants was less predictable. The high-frequency portion of the phonemes contains the most ‘information’ for the purposes of telling phonemes apart, and the removal of the low frequency formants leaves that information there.)

At the right side is a list of 16 consonants. With perfect hearing, all 16 sound different. As frequency loss chips away the unique parts of a phoneme, the blue lines are joined. Where they cross the slope is a yellow dot. Tracing from the yellow dot back to the phoneme list at the right, you can see how many different phonemes have been blurred into that indistinguishable sound. With even some hearing loss at 5000 Hz, some start to blend together: see that /k/ and /p/ sound the same to someone who has even a small loss at 5000 Hz (the farthest-right slope). But /t/ still sounds distinct: its blue line crosses that right-most slope unconfused. But by the second slope, with a hearing loss starting at 2500 Hz, /t/ is now confused with /p/ and /k/.

If your audiogram matches the farthest-left slope, with some loss as low as 300 Hz and a total loss by 4000 Hz, you would hear only three distinct consonant, one that could be /t/, /k/, /p/, /f/, /th/, /s/, or /sh/, another that could be either /m/ or /n/ and a third that could be /d/, /g/, /b/, /v/, /th/ (as in than), /z/, or /zh/. Fortunately, many of the confusable consonants are easily discriminated by lipreading: /k/ is made at the back of the mouth while /p/ is on the lips. However, the farther left the slope is, the more impossible it is to ‘hear’ on the phone, in dim light, with facial hair, etc.