Cordes A K, Ingham R J, Frank P, Ingham J C
Department of Speech and Hearing Sciences, University of California, Santa Barbara 93106.
J Speech Hear Res. 1992 Jun;35(3):483-94. doi: 10.1044/jshr.3503.483.
In response to the recognized need for a valid and reliable way to measure stuttering, this study investigates a measurement methodology based on time-interval analyses of stuttering event judgments. Three groups of judges, differing in stuttering judgment experience, identified stuttering events in 12 repeated presentations of five 1-min speech samples. Fixed time intervals ranging from 0.2 sec to 7.0 sec were then superimposed on the event judgments by a data analysis program. Inter- and intrajudge interval-by-interval agreement, and agreement for total numbers of intervals containing stuttering event judgments, were calculated for each judge group. Results showed that agreement was superior among more experienced judges and in longer interval lengths. Agreement varied across speech samples but not across the repeated judgment opportunities. Agreement was maximized at greater than chance levels for an interval of approximately 3.0 sec, but even this best agreement did not exceed a mean of approximately 60% for experienced judges.
为了满足对一种有效且可靠的口吃测量方法的公认需求，本研究调查了一种基于口吃事件判断的时间间隔分析的测量方法。三组在口吃判断经验上有所不同的评判员，在五个1分钟语音样本的12次重复呈现中识别口吃事件。然后，一个数据分析程序将从0.2秒到7.0秒的固定时间间隔叠加到事件判断上。计算了每个评判员组的评判员之间和评判员内部逐间隔的一致性，以及包含口吃事件判断的间隔总数的一致性。结果表明，经验更丰富的评判员之间以及较长的时间间隔内一致性更好。一致性在不同语音样本间有所变化，但在重复的判断机会中没有变化。对于大约3.0秒的间隔，一致性在大于随机水平时达到最大值，但即使是这种最佳一致性，对于经验丰富的评判员来说，平均也不超过约60%。
