Examqo › Examqo library › Speech-language pathology and audiology › Clinical audiology
Speech audiometry: SRT, SDT, word recognition, rollover and dynamic range
Speech audiometry tests three things — the level at which speech is detected (SDT), the level at which half of two-syllable words are understood (SRT), and how many words are understood at a comfortable level (word recognition) — and it is also used to check consistency with the audiogram and to raise suspicion of a retrocochlear lesion or a functional loss.
Why it matters in the exam
Questions test definitions and relations: the SRT is tested with two-syllable words with equal stress (spondees) and is the level at which the listener repeats 50% of the words; the SDT marks the softest level where speech is noticed on half the trials, and it is about 10 dB better than the SRT; a normal SRT agrees with the pure-tone average — an SRT much better than the average raises suspicion of a functional loss. In the psychometric function the x-axis is level in dB and the y-axis is percent correct, and with two-syllable words 100% is reached already 5–10 dB above threshold.
The second part is interpretation: a drop in word recognition as the level is raised (rollover) supports a retrocochlear lesion; 60% recognition at 35 dB above threshold in someone with a mild loss raises suspicion of a retrocochlear lesion; 96% at only 10 dB above threshold raises suspicion of a functional loss; and with a flat audiogram, testing 35 dB above the SRT gives full access to speech sounds. The dynamic range is the distance between threshold and the uncomfortable loudness level (UCL), and it is narrow in cochlear loss because of recruitment. A person who complains he cannot hear in noise is tested with speech in babble; and older adults' difficulty understanding fast speech comes mainly from poor temporal resolution.
Key points
- SDT — detection; SRT — 50% of two-syllable words; WRS — percent understood.
- The SRT agrees with the pure-tone average; a much better SRT — suspect functional loss.
- Rollover and lower-than-expected recognition — suspect retrocochlear.
- Dynamic range = UCL minus threshold; narrow with recruitment.
- Speech in noise for noise complaints; older adults — temporal resolution.
Summary table
| Test | What is tested | Normal value or interpretation |
|---|---|---|
| SDT | The level where the listener notices speech on half the trials | About 10 dB better than the SRT; slightly better than the best audiogram threshold |
| SRT | The level at which 50% of two-syllable words (spondees) are repeated | Agrees with the pure-tone average within about 10 dB; a much better SRT — suspect a functional loss |
| Word recognition (WRS) | Percent of one-syllable words repeated correctly at a comfortable level | Lower than expected for the degree of loss — suspect retrocochlear; constant masking level |
| Rollover | A fall in recognition as the level rises beyond the peak | Positive — supports a retrocochlear lesion; tested with one-syllable lists at rising levels |
| UCL and dynamic range | The uncomfortable level; the distance between it and threshold | A narrow range in cochlear loss (recruitment) |
| Speech in noise | Recognition in babble noise or in sentences (for example AzBio) | Suits a complaint about hearing in noise; older adults do worse than the audiogram predicts |
In the full version for subscribers
- A simple algorithm
- Memory aid
- Exam traps
- 5 review cards
- What matters for the exam
The full version is open to subscribers of: Speech and hearing (communication disorders).
Subscriber? Read the full topic Plans and prices 25 free questions
Sources
- Katz (ed.), Handbook of Clinical Audiology, 7th edition (2015) — Speech audiometry
- ASHA Guidelines, 1988 — Guidelines for Determining Threshold Level for Speech
- Jerger & Jerger, Archives of Otolaryngology, 1971 — Diagnostic significance of PB word functions (rollover)
Sources are given for further reading; the text is written afresh and not copied from them.
Reviewed by the Examqo AI Professor · 2026-11-02. The text may contain errors; when in doubt, check the source.