Speech Audiometry

Speech audiometry tests how well a person hears and understands speech. Pure tone audiometry shows what the quietest sounds are that can be heard, but it does not show how clearly speech is understood, and the two can differ greatly. People with the same pure tone thresholds may have very different difficulty in daily conversation, particularly in noise.

Two main measures are used. The speech recognition threshold (or speech reception threshold) is the lowest level at which the person can repeat 50% of simple two-syllable words (spondees). The word recognition score (speech discrimination) is the percentage of single-syllable words that the person repeats correctly when they are presented at a comfortable, clearly audible level (usually 30 to 40 dB above the threshold).

Word lists are phonetically balanced, recorded in the patient's language, and presented through headphones or loudspeakers, in quiet. Additional tests include speech in noise, sentence tests, and the most comfortable and uncomfortable loudness levels, used for hearing aid fitting.

The test takes 10 to 20 minutes. The speech threshold should agree with the pure tone average within about 10 dB, which makes it a check on the reliability of pure tone thresholds. In normal hearing and in conductive hearing loss, word recognition scores are 90 to 100%. In sensorineural hearing loss, scores are lower, and in lesions of the auditory nerve and brainstem, they may be very poor.

Indications.

• Everyone with a hearing complaint, together with pure tone audiometry, to find the real hearing handicap.
• Confirmation of pure tone thresholds, and detection of unreliable or non-organic responses.
• Assessment of hearing aid candidates, and hearing aid performance, and cochlear implant candidacy.
• Suspected retrocochlear disease: asymmetric loss with poor speech discrimination that is out of proportion to pure tone loss, such as vestibular schwannoma.
• Auditory neuropathy and central auditory processing problems, with difficulty hearing in noise.
• Monitoring of change in hearing after sudden loss, surgery, treatment, or ototoxic exposure.
• Medico-legal and disability assessment.
• Difficulty understanding speech in noise despite normal pure tone thresholds. Not suitable for patients who cannot understand or repeat the words, such as infants or those with language problems, for whom picture-pointing tests or other measures are used.

Contraindications and precautions.

• No contraindications.
• The test language and word lists must match the patient's language and dialect. Interpreters, or lists in the home language, are used. Differences in vocabulary and accent change scores.
• Patients with speech or cognitive problems, learning disability, or dementia may not be able to repeat the words, and adapted tests are used.
• Ear problems (wax, infection) are treated or noted.
• A good listening environment and calibrated equipment are needed, and the test level is set properly.
• Masking noise is used in the non-test ear when the test ear is much worse than the other, to avoid cross-hearing.
• Very high presentation levels can be uncomfortable or damaging, and levels are controlled.

Preparation and technique.

Pure tone audiometry is done first, and the pure tone average is calculated. The patient is told to repeat each word that they hear. For the speech recognition threshold, two-syllable words are presented starting at a level that is clearly heard, and lowered in steps until the patient repeats about half of them correctly.

For the word recognition score, a list of 25 or 50 single-syllable words is presented at a comfortable level (often 30 to 40 dB above the speech threshold, or at the most comfortable level), and the percentage correct is calculated. The test is repeated at several higher levels when rollover (a fall in score with increasing level) is suspected. Each ear is tested separately, with masking as needed. Speech in noise tests are given with background noise at a set signal-to-noise ratio.

Normal appearances.

The speech recognition threshold is within about 10 dB of the pure tone average, and is 0 to 20 dB HL in a person with normal hearing. The word recognition score is 90 to 100% at a comfortable level. The performance-intensity curve rises steeply to a plateau, with no fall at higher levels. In background noise, normal listeners understand speech at a signal-to-noise ratio of about 0 to 2 dB or better.

Interpretation.

• Conductive hearing loss: the speech threshold is raised in line with the pure tone loss, and word recognition is normal (90 to 100%) once the sounds are made loud enough.
• Cochlear sensorineural loss: reduced word recognition that depends on the degree and type of loss. Mild to moderate losses give scores of about 70 to 90%, moderately severe losses give 50 to 70%, and severe to profound losses give lower scores. The scores are lower in presbycusis with steeply sloping loss, and in noise-induced loss.
• Retrocochlear disease (vestibular schwannoma, auditory nerve and brainstem lesions): poor word recognition that is out of proportion to the pure tone loss, and rollover (a fall in score of over 20 to 25% when the level is raised). A difference in scores between the two ears of more than 15 to 20% raises suspicion.
• Auditory neuropathy: poor speech understanding that is worse than expected from the pure tone thresholds, and a poor response in noise.
• Central auditory processing disorders: normal scores in quiet, and poor scores in noise or with competing speech.
• Disagreement between the speech threshold and the pure tone average of over 10 to 15 dB points to unreliable results, non-organic loss, or an error.
• Hearing aid use: aided scores show the benefit, and a very low score with a hearing aid suggests that the aid is not helping enough and that a cochlear implant may be considered.
• Results are combined with tympanometry, reflex testing and imaging.

Critical values and findings needing immediate action.

• Asymmetric poor word recognition, or marked rollover: an MRI of the internal auditory canals for a vestibular schwannoma or other retrocochlear lesion.
• A sudden fall in speech understanding with sudden hearing loss: urgent assessment on the same day, treated as sudden sensorineural loss.
• Poor speech understanding with neurological signs (facial weakness, vertigo, imbalance, headache): urgent assessment.
• A sharp fall in scores in a patient on ototoxic drugs: urgent review of treatment.
• Poor aided scores in a child or adult: referral for implant assessment.

Limitations and artefacts.

• Results depend on the language, dialect, word lists, and the speaker (live voice or recorded). Live voice testing varies between examiners, so recorded material is better.
• Familiarity with the words, education, vocabulary, attention, memory, and cognitive function affect the scores, and not only hearing.
• The test in quiet does not reflect real-life conditions with noise, reverberation, and several speakers, and speech in noise tests are more sensitive for everyday problems.
• The score at a single level may miss rollover, and testing at one level can give a falsely normal result in retrocochlear disease.
• Practice effects, guessing, and lip reading (the examiner's face should be covered) change results.
• Word lists have different difficulty, and recorded lists need local validation.
• A normal score does not exclude a retrocochlear lesion, and imaging is based on the clinical picture.

Special groups.

• Children: picture-based and age-appropriate word lists, with play. Language and development affect scores.
• Elderly patients: scores fall with age, partly from changes in the brain and cognitive function, and not only in the ear.
• People who speak several languages, or have a different first language: testing in the language they use daily.
• Patients with hearing aids and implants: test both unaided and aided, and in noise.
• Patients with auditory neuropathy: scores are poor with sound that is audible, and amplification may help little.
• Regions: lists in local languages are often not available, and testing may be done with live voice lists developed locally, with care to use the same speaker, volume and method. Training of staff, and recorded lists in major languages, improve quality. Speech tests help decide who benefits from hearing aids, which are expensive and scarce.

Follow-up tests.

• Pure tone audiometry, tympanometry, and acoustic reflex testing.
• Auditory brainstem response for retrocochlear lesions.
• MRI of the internal auditory canals and brain.
• Speech in noise testing, and auditory processing assessment.
• Hearing aid fitting and review, and assessment for cochlear implantation.
• Cognitive assessment in older patients with poor scores.
• Repeat speech audiometry to monitor change.