December 17, 2011

Factors that enhance Australian speech-language pathologists’ assessment of the speech of Cantonese-speaking children

Rebekah Lockart has finished her Masters thesis titled: Factors that enhance Australian speech-language pathologists’ assessment of the speech of Cantonese-speaking children. It has been an honour to supervise her project as part of her Master of Speech and Language Pathology in the Faculty of Human Sciences, Department of Linguistics at Macquarie University, Sydney. Here is her thesis abstract:

The aim of this study was to investigate the ability of 33 Australian speech-language pathology (SLP) students to identify and transcribe typical and atypical speech in a nonnative language. Participants completed 3 tasks in an experimental within subjects design. Task 1 involved transcription of 5 English words to provide a baseline of their transcription skills. In Task 2 participants transcribed a typical Cantonese-speaking adult from an audio recording of 25 words from the Hong Kong Cantonese Articulation Test (HKCAT). The listeners transcribed an average of 59.1% consonants correctly. The participants’ average score was increased to 72.9% when a transcription scoring system was applied (2=exact match, 1=common transfer pattern, 0=incorrect). In Task 3 participants were presented with 100 audio-visual recordings of Cantonese-speaking children producing words from the HKCAT and a phonetic transcription of each word. Participants identified consonant speech sound errors and transcribed each word under 4 additive conditions: 1) baseline, 2) +recording of an adult model, 3) +information about the Cantonese phonological system, 4) all variables. In Condition 1 participants accurately identified an average of 63.8% of children’s whole word productions as correct or incorrect. Participants achieved an average transcription score of 71.2%. The accuracy of speech sound error identification and transcription was significantly improved by the provision of information about the Cantonese phonological system (69.2%, 76.1%), and further enhanced by a recording of an adult model (71.6%, 76.1%), and addition of both factors (72.8%, 79.8%). Accuracy was influenced by broad transcription skill and proficiency in LOTEs, but not by musicality or confidence in working with multilingual clients. These results indicate SLP students, with no exposure to or specific training in Cantonese, have some ability to transcribe the speech of Cantonese-speaking adults and children and identify speech sound errors made by Cantonese-speaking children.