Purpose The goal of this study was to determine whether the results obtained from a 25-utterance conversational language sample were as reliable as those obtained from a 50-utterance sample. Method Robust conversational language samples from 220 children with typically developing language (106 boys, 114 girls) ranging in age from 3;2 to 7;10 (years;months) were collected. The language samples were randomly assigned to one of two conditions: a 25-utterance condition and a 50-utterance condition. Transcripts were examined for three metrics, including mean length of utteranceSUGAR, words per sentence, and clauses per sentence. Results Data were analyzed using two methods. A linear mixed-model analysis was used to assess absolute and relative reliability, and the Bland-Altman procedure was used to assess absolute reliability and clinical acceptability. Results of the mixed-model analysis indicated that mean length of utteranceSUGAR and words per sentence demonstrated relative reliability; however, none of the metrics demonstrated absolute reliability. In contrast, results of the Bland-Altman scatter plots indicated that all three metrics demonstrated absolute reliability because 94%-96% of participants' scores fell within the limits of agreement. Taken together, the results suggested that the statistically significant differences indicated by the mixed-model analysis were not clinically significant. Conclusion These results highlighted the importance of using different methods of analysis in studies of reliability. The findings indicated that reliable language sample results can be obtained from 25-utterance samples. Furthermore, by including practices already in use (e.g., collecting samples ≤ 50 utterances) and including only minimal changes to current practices, the methods used in this study are feasible for school-based clinicians, could be easily integrated into clinical practice, and could increase the use of evidence-based assessment practices in schools.

中文翻译：

---

重新审视可靠性：使用抽样话语和经修订的语法分析（SUGAR）比较25种语言和50种语言的语言样本。

目的这项研究的目的是确定从25言语会话语言样本中获得的结果是否与从50言语样本中获得的结果一样可靠。方法收集了220名典型发展语言的儿童（106名男孩，114名女孩）的健壮对话语言样本，年龄从3; 2到7; 10（年；月）不等。语言样本被随机分配给以下两个条件之一：25语音条件和50语音条件。检查了笔录的三个指标，包括平均说话长度，糖，每个句子的单词和每个句子的从句。结果数据采用两种方法进行分析。线性混合模型分析用于评估绝对和相对可靠性，而Bland-Altman程序用于评估绝对可靠性和临床可接受性。混合模型分析的结果表明，话语糖的平均长度和每个句子的单词显示出相对的可靠性。但是，没有任何一项指标显示出绝对的可靠性。相反，Bland-Altman散点图的结果表明，所有这三个指标都显示出绝对的可靠性，因为94％-96％的参与者的评分落在协议的范围内。两者合计，结果表明，混合模型分析表明的统计学上的显着差异在临床上不显着。结论这些结果强调了在可靠性研究中使用不同分析方法的重要性。这些发现表明，可以从25个发音样本中获得可靠的语言样本结果。此外，通过包括已经使用的实践（例如，