Department of Psychology, Carnegie Mellon University, Pittsburgh, PA.
Department of Neurosurgery, Baylor College of Medicine, Houston, TX.
J Speech Lang Hear Res. 2021 Apr 14;64(4):1271-1282. doi: 10.1044/2020_JSLHR-20-00561. Epub 2021 Mar 30.
Purpose Analysis of connected speech in the field of adult neurogenic communication disorders is essential for research and clinical purposes, yet time and expertise are often cited as limiting factors. The purpose of this project was to create and evaluate an automated program to score and compute the measures from the Quantitative Production Analysis (QPA), an objective and systematic approach for measuring morphological and structural features of connected speech. Method The QPA was used to analyze transcripts of Cinderella stories from 109 individuals with acute-subacute left hemisphere stroke. Regression slopes and residuals were used to compare the results of manual scoring and automated scoring using the newly developed C-QPA command in CLAN, a set of programs for automatic analysis of language samples. Results The C-QPA command produced two spreadsheet outputs: an analysis spreadsheet with scores for each utterance in the language sample, and a summary spreadsheet with 18 score totals from the analysis spreadsheet and an additional 15 measures derived from those totals. Linear regression analysis revealed that 32 of the 33 measures had good agreement; was the one score that did not have good agreement. Conclusions The C-QPA command can be used to perform automated analyses of language transcripts, saving time and training and providing reliable and valid quantification of connected speech. Transcribing in CHAT, the CLAN editor, also streamlined the process of transcript preparation for QPA and allowed for precise linking of media files to language transcripts for temporal analyses.
目的
分析成人神经源性语言障碍领域的连续言语对于研究和临床目的至关重要，但时间和专业知识通常被认为是限制因素。本项目的目的是创建和评估一个自动程序，以对定量生产分析（QPA）进行评分并计算其度量值，这是一种客观而系统的方法，用于测量连续言语的形态和结构特征。
方法
使用 QPA 分析了 109 名急性亚急性左侧大脑半球中风患者的灰姑娘故事转录本。使用 CLAN 中的新开发的 C-QPA 命令（一组语言样本自动分析程序），使用回归斜率和残差比较手动评分和自动评分的结果。
结果
C-QPA 命令生成了两个电子表格输出：一个分析电子表格，其中包含语言样本中每个话语的分数，以及一个汇总电子表格，其中包含分析电子表格中的 18 个分数总和以及从这些总和中得出的另外 15 个度量。线性回归分析表明，33 个测量值中有 32 个具有良好的一致性；有一个得分没有良好的一致性。
结论
C-QPA 命令可用于执行语言转录本的自动分析，节省时间和培训，并提供可靠和有效的连续言语量化。在 CLAN 编辑器 CHAT 中进行转录也简化了 QPA 的转录准备过程，并允许将媒体文件与语言转录本精确链接，以便进行时间分析。
