Word Recognition – Speech Testing With Dr René Hornby

Speech is the auditory stimulus through which we communicate. Auditory assessment of hearing loss using speech is therefore of great interest in the field of speech and hearing sciences; in fact, it has a long history in the evaluation of hearing from as early as the 1800s.

The hearing test using speech audiometry has been studied as early as 1804, but the validity and reliability of speech testing is as important today as it was then.

Dr René Hornby, a leading researcher and clinician, is our guest this month for the Q&A. She has a great passion for ensuring valid and reliable speech testing specific to the patients context.

eMoyo is pleased to bring Dr Hornby’s knowledge and clinical expertise in the area of speech testing to the Q&A.

This excellent interview offers information that will be helpful to you as the clinician to best implement evidence-based practice in speech audiometry testing in hearing health care. 

René Hornby, B.Logopedics & M.Communication Sciences (UP), Rem dipl (RAU), Doctor of Audiology (A.T. Still, USA), is a clinical audiologist and owner of Dr. René Hornby Audiologists in Pretoria, South Africa.

She has vast interest in clinical and research audiology including speech perception and recognition in patients with both normal and impaired hearing.

Dr. Hornby has a passion for ensuring ‘healthcare for all’, and one of her solutions, together with her team, is creating, re-/recording and validating speech material to ensure contextualised and reliable speech audiometry assessment and management.

Why should clinicians conduct speech audiometry?

They can get all that information from pure-tone thresholds, right?

Dr René Hornby (RH):

Wrong. Speech audiometry has been part of the basic audiometric test battery for the past 70 years, since the first development of the PAL PB 50 in 1948.

The reason for developing speech test stimuli was that pure tone thresholds do not provide us with information on how a person perceives speech and/or how the loss of hearing sensitivity impacts on speech perception.

Back then researchers claimed that speech audiometry results would provide us with a better understanding of how a patient is communicating. We now know that this is not entirely accurate or realistic. Listening to words in quiet does not tell us how a person is hearing or communicating in the real world (Hornsby & Mueller, 2015).

Oh, I see.

With that said, do you think speech recognition tests add any clinical value by being part of the audiology test battery?


Most certainly. It helps quantify benefit from amplification and assists in determining programming and audiologic management decisions (e.g., consideration of alternative devices, FM, need for additional classroom supports).

For example, when presenting lists like the AB (Arthur Boothroyd) word lists which are isophonemic, we can determine which speech sounds are really inaudible or difficult to hear for the patient, since the same 10 vowels and 20 consonants are repeated in each list.

When the patient has been fitted with hearing aids, and we repeat these AB lists, (never the same list!) we can determine whether the vowels and consonants are audible in a controlled test environment since the words are RECORDED and validated (Wilson et al, 1998).

So, unaided and aided scores are compared with each other. Word recognition results can be used successfully to monitor performance over time BUT only if RECORDED material is used. It can also add value when comparing unaided and aided scores with each other. Remember, the greatest complaint of people with a hearing loss is difficulty hearing in noise.

We recommend conducting word recognition in noise (using a multitalker babble noise) with a +5 dB and +10 dB S/N unaided and/or under aided conditions. Speech recognition testing may prove useful in detecting Auditory Neuropathy Spectrum Disorder (ANSD), especially testing in noise (Hillock-Dunn, 2015). The results are adding value especially for counselling and management purposes.

On that note, in your professional opinion, can speech audiometry results be used as a diagnostic tool to rule out certain pathologies?


Yes and no. Back in the 70’s and 80’s word recognition scores (aka speech discrimination) were used to diagnose cochlear and retrocochlear pathologies (observing the roll-over in patients with acoustic neuromas, and a plateau in cochlear losses).

“In ancient history (early 1970s) we did a PI/PB function and pretended it had sensitivity/specificity (it didn’t), that was before the routine use of acoustic reflexes, OAEs and MRI” (Hornsby & Mueller, 2013). According to James Hall (2017), SRT and word recognition testing in QUIET does not add diagnostic value to the standard test battery.

Keep in mind that TODAY we have more sensitive and specific test measures to identify cochlear and retrocochlear pathologies. Speech audiometry remains a subjective measure.

So no, we would not diagnose a pathology based on only subjective test results like speech audiometry, but would rather depend on our results of OAEs and acoustic reflexes.

BUT we would be concerned if a patient presents with poor word recognition abilities/score especially lower than or equal to 50/52% or if the patient’s WR score is disproportionately low compared to the PTA (as seen on the SPRINT chart when using the NU 6 lists).

Reason for concern should also be significantly different scores (according to the binomial model) between two ears when the degree of loss is symmetrical.

What are the clinical functions of Speech Reception Threshold and Word Recognition measures?


The value of SRT is its use as a cross-check principle (in order to determine whether the PTA and SRT correspond) – this is particularly useful when testing children and malingerers.

The goal of word recognition testing (previously referred to as speech discrimination testing) is to determine the patient’s optimum performance for word recognition under controlled and standardised conditions (Hornsby & Mueller, 2015).

You can ask yourself: “How good does it get for the patient?” / Pbmax.

Word recognition has been, and still is, being referred to as speech discrimination.

Can these terms be used interchangeably?


No. According to James Hall, we have been using and applying the term speech discrimination incorrectly. You will notice that the term ‘word recognition’ (and that is what we are testing) has been used in textbooks since the 90’s.

Do you still remember that in Audiology 101 we were taught Erber’s auditory hierarchy? To recognise and repeat words is a higher level function than that of discrimination.

So, all audiologists have to use the correct term on their PI/PB curve, in their reports and when conversing with colleagues and other professionals.

I know some clinicians may not understand why a certain word list is used.

How do we know which word list/speech materials to use for a particular population/patient?


Quite correct, there are so many lists and copies of lists out there. Some do not even have a NAME and origin, which makes them invalid. Follow these guidelines when deciding which lists to use for your patients:

  • It must be recorded according to the latest industry standards. Remember the word list is NOT the test, BUT the recording of it is.
  • The lists should be phonetically balanced, which means that the speech sounds in the list are representative of the frequency of occurrence of those sounds in the spoken language. For example, the FVEWA (Foneties verteenwoordigende eenlettergrepige woordlyste in Afrikaans) consists of six phonetically balanced word lists which have recently been standardised (Naude, 2018).
  • The more test items in a list, the better statistical values can be obtained. We know nobody wants to conduct a 50-word list, but no need to panic. Good options to purchase are the NU 6 II (ordered by difficulty) (Hurley & Sells, 2003) and the Arthur Boothroyd word list when scoring phonemically.
  • The FVEWA can be used for Afrikaans first language speakers from 5 years old and upwards and the AB word lists can be used for English first language speakers from 5 years old and upwards (McCreery et al, 2010).
  • The recording should be of a native first language presenter.
  • Don’t choose recorded lists with foreign accents. Several researchers have proven that the patients must be evaluated in their first language to ensure the test validity and reliability (Roets, 2005).

Why can’t we all use the same international word lists of the same language for word recognition test or speech audiometry?


  • The recording should be of a native first language presenter.
  • Don’t choose recorded lists with foreign accents. Several researchers have proven that the patients must be evaluated in their first language to ensure the test validity and reliability (Roets, 2005).

Right, thank you. So can I use monitored live voice (MLV) to present the speech materials?

They are much faster and flexible to present.


Yes you may use live voice when performing SRT measurements but NOT for recognition purposes, whether it be words, sentences, in quiet or in noise.

STOP the MADNESS stated Ross Roeser and Jackie Clark already in 2008. When using live voice, your test reliability and validity of results are significantly reduced.

And NO, it takes only 1 min 18 seconds to present a recorded 25-word list versus 1 min 2 sec when using live voice. When using recordings you have applied BEST practice (HPCSA, 2002), your conscience is clear and you may sleep peacefully.

Have a look at this article to discover how a single device could boost your practice.


  • Dubno, JR et al (1995). Confidence limits for maximum word-recognition scores. Journal of Speech & Hearing Research. Volume 38, 490 – 503, April.
  • Hall. W, James III (2017). Rethinking the Audiology Test Battery. Audiologyonline.
  • Hillock-Dunn, A (2015). Pediatric Speech Recognition measures: What’s now and what’s next! Audiologyonlie.com/20Q.
  • Hornsby, B & Mueller, HG (2013). Monosyllabic Word Testing: Five simple steps to improve accuracy and efficiency. www.audiologyonline.com (Course no 22930 & 22139), www.audiologyonline.com.
  • HPCSA, 2002. Standards of Practice in Audiology. Professional Board for Speech, Language and Hearing Professions.
  • Hurley RM & Sells, JP, (2003). Ear & Hearing. An abbreviated word recognition protocol based on item difficulty.
  • Kramer, S (2008). Audiology: Science to Practice.
  • Leslie A Guthrie & Carol L Mackersie, (2009). A comparison of presentation levels to maximize word recognition scores. JAAA.
  • McCreery, R et al (2010). Performance-intensity function for normal-hearing adults and children using the CASPA, Ear & Hearing.
  • Mueller, G & Hall, JW (1997). Audiology desk reference, Audiology Diagnosis.
  • Myles, AJ (2017). The clinical use of Arthur Boothroyd (AB) words lists in Australia: exploring evidence-based practice. International Journal of Audiology, Early online: 1-6.
  • Naude, A (2018). Statistiese evaluering en voorspelling van woordherkenning-toetstellings van die Foneties Verteenwoordigende Eenlettergrepige Woordlyste in Afrikaans (FVEWA). Tydskrif vir Geesteswetenskappe, 58(1) bl 124-136.
  • Roeser, RJ & Clark, JL (2008). Live voice speech recognition audiometry: STOP the madness! Audiology Today p 32-33.
  • Roets, R (2005). Spraakoudiometrie in Suid-Afrika: Ideale kriteria teenoor kliniese praktyk. UP. Unpublished dissertation.
  • Schoepflin, JR (2012). Back to Basic: Speech Audiometry. AudiologyOnline.
  • Textbook: Gelfand, SA (2006). Essentials of Audiology.
  • Wilson, W., Jones, B & Fridjohn, P (1998). Use of the NAL AB wordlists as a South African English Speech Discrimination Test. South African Journal of Communication Disorders, 45, 77-86.