Should I use dB HL for speech audiometry?

The presentation level of speech stimuli is sometimes expressed in dB SPL, sometimes in dB HL.  

In principle, dB SPL is the correct physical scale.  The sound level is measured with a sonometer and expressed in dB SPL. However, two words with the same intensity are not necessarily perceived by the ear as being equally loud.  This is because the ear's sensitivity to different frequencies is different.  So if a word contains mainly frequencies around 2000 Hz, 60 dB SPL will sound louder than for a word with mainly low frequencies.  Some speech lists compensate slightly for these differences, to avoid the subjective loudness having too great an influence on the score.

In any case, the score on a speech list will improve as the presentation intensity increases, which is expressed by the typically sigmoidal norm curve.

Each speech list has its normal curve.  In this example, a score of 50% is achieved at a presentation level of 23 dB SPL. 

By analogy with audiometry, here too we can define 0 dB HL as the presentation level at which 50% of the responses are correct, in this case, 0 dB HL = 23 dB SPL.  We can indicate this graphically by the second horizontal axis.  In Audiqueen we denote the dimension as dB HL!  The exclamation point serves to indicate that this is specific to this one speech list. 

Audiqueen automatically converts dB SPL to dB HL! if you check the corresponding box in the format window:

Audiqueen is now configured to show dB SPL as the primary horizontal axis
and a stimulus-specific dB HL! as the secondary axis.

It is important to note that this dB HL! scale is different for each speech list.  This means that e.g. an SRT score of 53 dB HL! corresponds to 70 dB SPL in speech test A, but 74 dB SPL for speech list B.

Several centres choose to use the dB HL! scale to be independent of the speech list.  They define a fixed offset of e.g. 20 dB or 23 dB between their own dB HL scale and the physical dB SPL scale.  This does not really make sense, but it is a common practice.  Audiqueen also allows this in the format window.

Audiqueen is now configured to show dB SPL as the primary horizontal axis
and a fixed-offset dB HL! as the secondary axis.

See also:

• Audiqueen's support page on decibels