Parrot evaluation

Reproduced from the British Society of Audiology News, February 1997. This work was performed with a very early version of the Parrot.

The Parrot Speech Discrimination Tester

An aid to screening for hearing impairment in the paediatric population


The use of speech discrimination tests in the Paediatric population above the age of two years, can confirm or raise suspicions of a hearing loss when other test methods prove inadequate or are simply unavailable. One of these test methods, the “McCormick Toy Discrimination Test” (McCormick,1977) has been developed into an Automated version, the “IHR- McCormick Automated Toy Discrimination Test (IHRATT)” (Ousey et al., 1989; Palmer et al., 1991; Summerfield et al., 1994). Predictions of better ear hearing thresholds can be made using this system which presents speech stimuli via a speaker and uses an algorithm to vary presentation level. A fairly recent addition to the equipment available for speech testing is the PARROT.

The PARROT is a portable digital speech screening system, consisting of

  1. A loudspeaker
  2. A boxed McCormick Toy Test
  3. A handset which enables the user to choose both the test item and the presentation level (40,50,60,70,80 dB(A) options).

The system runs off a rechargeable battery and comes complete with recharger/mains power supply. Digitised speech stimuli are used with a male speaker providing the speech samples.

As with the live voice version of the test, the objective is to identify the pass level at which the child can correctly identify 4 out of 5 items. At the lowest level, 40 dB(A), a child with normal hearing (better ear) should be able to identify word pairs with an 80% accuracy.

At the Children’s Hearing Assessment Centre, Nottingham we have been running an initial evaluation of this system and so far we have tested over fifty children. The objective so far has been to evaluate the PARROT as a screening device.


Fifty children above the age of 2 years were entered into the study and all attended the Children’s Hearing Assessment Centre (CHAC) for routine audiological assessment. Pure tone thresholds were recorded bilaterally using headphones, frequencies 0.5, 1, 2 and 4 kHz were tested. If pure tone audiometry was not successful, sound field thresholds were obtained using warble tone generators. The IHRATT then followed, and finally the PARROT was used to determine the quietest level at which 80% correct word identification took place. Background noise in clinic was typically around 18 dB(A).


As the PARROT is likely to be used as a screening device, the data was analysed in terms of;

  1. Sensitivity and specificity of identifying children who would score greater than 40 dB(A) using the IHR Automated Toy Test.
  2. Sensitivity and specificity of identifying children with averaged PTA thresholds (better ear) greater than 25 dB HL.

Summerfield et al. (1994) showed that an IHRATT score of 40 dB(A) equates to an average PTA threshold of 25 dB HL (better ear). Thus, theoretically the above two analyses were expected to yield similar values. Using the criteria given in (1) above, the sensitivity of the PARROT was 100% assuming we are trying to detect children who would score greater than 40 dB(A) on the IHRATT. The specificity was 86% which suggests that at least 4 out of 5 non-impaired children (as defined by the IHRATT score greater than, or equal to 40 dB(A) ) would be identified as such by the PARROT. Using the criteria given in (2) above, the sensitivity of the PARROT was similarly 100% assuming this time that we are trying to detect children who have pure tone averages (better ear of greater than 25 dB HL. The specificity was 80% which suggests that 4 out of 5 children with pure tone average thresholds (better ear) less than, or equal to 25 dB HL will be correctly identified.

Discussion and Conclusion

What does this information tell us?

  1. A child with a better ear PTA average of greater than 25 dB HL will fail the PARROT screening test at a level of 40 dB(A).
  2. A child who would require a level of greater than 40 dB(A) to correctly identify items using the IHRATT will also fail the PARROT test at a level of 40 dB(A).
  3. The PARROT will correctly identify around 80% of the children (using a PASS level of 40 dB(A)) who would either achieve an IHRATT score less than, or equal to, 40 dB(A) or a better ear PTA average less than or equal to 25 dB HL.

This initial evaluation is very encouraging, however data were obtained in controlled clinical settings with highly trained audiological staff. A more detailed analysis of this data is underway and quantities such as Positive Predictive Value will be investigated in conjunction with various screening “cutoff” levels to determine the desired optimum.

Sensitivity/specificity values.
In addition, the need to evaluate the ease of use of this equipment in a health visitor setting is essential, given that it is in these settings that the PARROT is most likely to be used.

In summary, the PARROT was easy to use, straightforward to calibrate and certainly worthy of further ongoing trials.


Ousey,J., Shepherd, S.,Twomey, T., PaimerA. R. The IHR McCormick Automated Toy Discrimination Test description and initial evaluation. British journal of Audiological Scientist Audiology 1989; 23:245-9.

McCormick, B. The Toy Discrimination Test: an aid for screening the hearing of children above a mental age of two years. Public Health, London 1977; 91: 67-9

SummerField, Q., Paimer A. R., Foster J. R., Marshall, D. H.,Twomey T. Clinical evaluation and test-retest reliability 0f the IHR-McCormick Automated Toy Discrimination Test. British Journal of Audiology 1994

Paul Shaw
Audiological Scientist
Children’s Hearing Assessment Centre

British Society of Audiology,
80 Brighton Road

Telephone 0118 966 0622