Stockman I J, Woods D R, Tishman A
J Psycholinguist Res. 1981 Nov;10(6):593-617. doi: 10.1007/BF01067296.
Noncry, nonlanguage vocalizations were sampled from four female babies between the ages of 7 and 21 months. Three trained listeners' phonetic transcriptions of more than 1,000 vocalizations were compared for interjudge and intrajudge agreement. The amount of agreement varied with the child's age and the criterion of agreement. The tendency toward somewhat greater interjudge agreement in the older than in the younger sampled months was attributed to the possibility that the child's vocal output becomes more speechlike with increasing age. Using an identical segment match criterion, interjudge and intrajudge agreement rarely exceeded 60% of the total number of segment comparisons made at any age. A feature-match criterion applied just to oral stops yielded higher agreement than did the identical segment match criterion. The results underscore the importance of considering listener reliability in assessing the validity of auditory descriptions of early vocal behavior and have implications for the methods used to describe auditory impressions.
从4名7至21个月大的女婴中采集了非哭泣、非语言发声样本。比较了三名经过训练的听众对1000多个发声的语音转录,以评估评判者之间和评判者内部的一致性。一致性程度随儿童年龄和一致性标准而变化。与年龄较小的抽样月份相比,年龄较大的月份中评判者之间的一致性略高,这可能是因为随着年龄增长,儿童的发声输出变得更像言语。使用相同片段匹配标准,评判者之间和评判者内部的一致性在任何年龄都很少超过片段比较总数的60%。仅应用于口腔塞音的特征匹配标准比相同片段匹配标准产生了更高的一致性。结果强调了在评估早期发声行为的听觉描述的有效性时考虑听众可靠性的重要性,并对描述听觉印象的方法具有启示意义。