3Speech Audiometry

FFrom thresholds to function

Pure-tone audiometry tells you how faint a tone the ear can detect; it says little about how well the patient understands speech. Speech audiometry closes that gap. The speech recognition threshold (SRT) is the lowest level at which the patient correctly repeats simple two-syllable spondee words half the time; it should agree, within a few decibels, with the pure-tone average. A large discrepancy is a flag for non-organic loss or test error.

The more clinically informative measure is the word recognition score (WRS) — the percentage of monosyllabic words repeated correctly at a comfortable supra-threshold level. WRS captures the clarity of hearing, which is precisely what tinnitus patients struggle with and what amplification has to restore [2005].

TWord recognition and the cochlear–retrocochlear question

In ordinary cochlear hearing loss, word recognition tracks the audiogram: as thresholds worsen, scores fall in a predictable way. The diagnostic power of speech audiometry lies in detecting disproportion — a word recognition score far poorer than the pure-tone loss would predict. Such a mismatch points away from the cochlea and toward the auditory nerve or brainstem, the territory of retrocochlear disease such as vestibular schwannoma [2014].

This is most meaningful when it is asymmetric: a substantially worse word score in one ear, especially the ear with the louder or unilateral tinnitus, strengthens the case for MRI of the internal auditory canal. Symmetric, proportionate word scores in bilateral tinnitus are reassuring and consistent with the common cochlear picture [2013].

CRollover: the classic retrocochlear sign

If word recognition is tested not at a single level but across increasing intensities, the cochlear ear keeps its best score as the sound gets louder. A retrocochlear ear behaves differently: the score peaks, then paradoxically declines as intensity rises further. This decline is called rollover, and it reflects the abnormal fatigue (adaptation) of a compromised auditory nerve under sustained loud stimulation.

Rollover is quantified as a rollover index — the drop from the maximum score to the score at the highest level, divided by the maximum. A large index has historically been treated as suggestive of a retrocochlear lesion. In modern practice it is one corroborating thread alongside acoustic-reflex decay, ABR and, definitively, MRI; it is suggestive rather than diagnostic, but in the patient with unilateral tinnitus it adds weight to the decision to image [2014].

TSpeech in noise: the patient’s real-world complaint

Tinnitus patients very often report that their main difficulty is following speech in noisy rooms — a complaint that a quiet-room audiogram and a quiet-room word score can entirely miss. Speech-in-noise tests reproduce that environment. The QuickSIN, for example, presents sentences against four-talker babble at descending signal-to-noise ratios and yields an SNR loss: the extra signal-to-noise ratio, in decibels, that the patient needs compared with a normal listener [2004].

This number does three things in the tinnitus clinic. It validates a real disability that pure tones underestimate; it predicts who will benefit from directional microphones or remote-microphone technology; and it gives an objective baseline against which to judge amplification and counselling. In the work-up it complements rather than replaces PTA — pure tones localise the lesion, speech-in-noise quantifies the handicap [2005].