AudiCloze

Appropriate for age 5 and over

Purpose

AudiCloze measures the ability to use knowledge of language to infer the identity of words that have not been identified because of noise, reverberation, or a listening difficulty of some sort. It is intended to be used by audiologists to check whether children being assessed for auditory processing disorders might in fact have a language deficit as a contributing factor to any reported listening difficulty, and/or as an important real-life ability that might have been impacted by an auditory processing disorder.

Nature of Task

The test comprises sentences with high redundancy within each sentence, but in which one word has been removed from the sentence and replaced with a whistle. The sentence is presented at a comfortable level in quiet. The child’s (or adult’s) task is to state either the missing word, or repeat the whole sentence, including the missing word, whichever response method they find easier. The clinician’s role is to indicate whether the word inserted by the client is appropriate to the sentence. To assist the clinician in determining this, the test screen shows the word that was originally in the sentence (i.e. before deletion) plus a selection of other common responses, all of which are considered as acceptable responses based on responses frequently made by children or adults and that are as consistent with the rest of the sentence as the original word that has been removed. For any other responses, the clinician judges whether the response makes sense in the sentence, or does not. For responses that have a sensible semantic meaning, but which differ in tense or plurality from an otherwise correct response, the clinician marks the response as “partially correct”. The screen, which is visible to the clinician but not the client, is shown below. The current sentence is displayed, with the deleted word highlighted.

Scoring

Normative data exist for children (based on typically developing British children listening to the British-accented version) and for adults (based on typically developing Australian young adults listening to the Australian-accented version. Scoring is computer assisted, and at the end of the test, the individual’s score is displayed in several formats. These comprise percent correct, z-score (i.e. number of standard deviations above mean for that age), scaled score and percentile. All except the percentage correct are automatically corrected for age, based on the normative data built into the test.

Scientific Studies

AudiCloze has been used in research studies at Macquarie University, University of Melbourne and the University of Manchester, the results of which will appear in future scientific articles. The following graphs show some of the results.

Figure 1 shows the relationship between speech intelligibility scores (with more negative scores representing better performance) and language ability measured with AudiCloze.

Figure 1. Speech reception threshold in noise versus language ability as measured with AudiCloze. Red symbols represent people with English as a second language and blue symbols represent native English speaker. From Master’s thesis of Ponsuang Luengtaweekul (Macquarie University).

Figure 2 shows the association between AudiCloze and speech understanding in noise, as measured with the Test of Listening Difficulties-Universal (ToLD-U). The most plausible reason for the significant association is that language ability helps with speech understanding in noise, but it is also possible that superior speech understanding in noise helps a child develop superior language ability.

Figure 2. Speech reception threshold in noise, expressed as age-corrected z-scores versus the age-corrected language ability as measured by AudiCloze. From PhD thesis of Xuehan Zhou (University of Manchester and University of Melbourne).

Figure 3 shows how age-corrected scores on AudiCloze compare to age-corrected scores obtained with the CELF sentence repetition test, which is commonly used by speech pathologists as an indicator of language ability. The AudiCloze and CELF sentence repetition tasks share the need for the child to understand the words, form the concept that the sentence is conveying, and hence remember the sentence. AudiCloze additionally requires the child to use this knowledge to make an inference about the word that is missing from the sentence, something that is possible only by using knowledge of the syntax and semantics of the language.

Figure 3. Age-corrected scores on AudiCloze versus those on the standard CELF sentence repetition test, for typically developing children aged from 5 to12 years. From PhD thesis of Xuehan Zhou (University of Manchester and University of Melbourne).