Brinca Lilia, Batista Ana Paula, Tavares Ana Inês, Pinto Patrícia N, Araújo Lara
Laboratory of Audiology and Speech and Language Therapy, School of Health, University of Algarve, Faro, Portugal.
Service of Physical Medicine and Rehabilitation, Algarve Central Hospital, Faro, Portugal.
J Voice. 2015 Nov;29(6):776.e7-14. doi: 10.1016/j.jvoice.2015.01.007. Epub 2015 Mar 17.
The main objective of the present study was to investigate if the type of voice stimuli-sustained vowel, oral reading, and connected speech-results in good intrarater and interrater agreement/reliability.
A short-term panel study was performed.
Voice samples from 30 native European Portuguese speakers were used in the present study. The speech materials used were (1) the sustained vowel /a/, (2) oral reading of the European Portuguese version of "The Story of Arthur the Rat," and (3) connected speech. After an extensive training with textual and auditory anchors, the judges were asked to rate the severity of dysphonic voice stimuli using the phonation dimensions G, R, and B from the GRBAS scale. The voice samples were judged 6 months and 1 year after the training.
Intrarater agreement and reliability were generally very good for all the phonation dimensions and voice stimuli. The highest interrater reliability was obtained using the oral reading stimulus, particularly for phonation dimensions grade (G) and breathiness (B). Roughness (R) was the voice quality that was the most difficult to evaluate, leading to interrater unreliability in all voice quality ratings.
Extensive training using textual and auditory anchors and the use of anchors during the voice evaluations appear to be good methods for auditory-perceptual evaluation of dysphonic voices. The best results of interrater reliability were obtained when the oral reading stimulus was used. Breathiness appears to be a voice quality that is easier to evaluate than roughness.
本研究的主要目的是调查语音刺激的类型——持续元音、朗读和连贯言语——是否能带来良好的评分者内和评分者间一致性/可靠性。
进行了一项短期的小组研究。
本研究使用了30名欧洲葡萄牙语母语者的语音样本。使用的语音材料为：（1）持续元音/a/；（2）朗读欧洲葡萄牙语版的《老鼠亚瑟的故事》；（3）连贯言语。在经过大量文本和听觉锚定训练后，要求评判者使用GRBAS量表中的发声维度G、R和B对发声障碍语音刺激的严重程度进行评分。在训练6个月和1年后对语音样本进行评判。
对于所有发声维度和语音刺激，评分者内一致性和可靠性总体上非常好。使用朗读刺激获得了最高的评分者间可靠性，尤其是对于发声维度等级（G）和呼吸音（B）。粗糙度（R）是最难评估的语音质量，导致所有语音质量评分中评分者间不可靠。
使用文本和听觉锚定进行广泛训练以及在语音评估过程中使用锚定似乎是对发声障碍语音进行听觉-感知评估的好方法。使用朗读刺激时获得了评分者间可靠性的最佳结果。呼吸音似乎是比粗糙度更容易评估的语音质量。
