Status | Completely developed architecture with software for generating tests to evaluate intelligibility at the word level: item construction, test administration, and data analysis [Jekosch (1992)]. |
Goal | Diagnostic and comparative evaluation of the intelligibility |
of sequences of one or more consonants (``consonant clusters '') and sequences of one or more vowels (``vowel clusters''). | |
Languages | In principle any language, as long as matrices with permissible phoneme and phoneme combinations are available. |
Items | Mostly meaningless but also (by chance) meaningful, open vocabulary taking into account phonotactic constraints ; combinatory matrices are available for English, French, German, and Italian. The frequency of the syllable structures (e.g. CCVCC, VCC) and clusters to be included in the test material can be defined. Examples: storp and ast. |
Procedure | Open response identification using unambiguous notation or normal spelling. |
Time | With 900 stimulus words (presented by [Jekosch (1992)], about 2 hours per synthesiser. However, subsets of stimuli can also be defined. |
Analysis | The SOAP scoring module allows, among other things, the calculation of percentages correct of all clusters, of initial, medial, and final clusters separately, and of complete items. Confusion matrices are generated. |