Automated coding has the potential to streamline otherwise laborious manual coding of psychologically relevant human behaviour. For example, automated coding of adults’ written language (e.g., social media posts) shows that incidental language patterns are associated with internal psychological processes, including symptoms of depression. However, it is unclear whether automated coding of children’s incidental spoken language is related to their symptoms and other indices of risk. We therefore examined associations between 146 children’s (M_age = 9.53 years, SD = 0.88; 82 girls) spoken language, transcribed for automated coding, and their internalizing symptoms and cortisol stress reactivity. Children completed self-reports of depression and anxiety, and cortisol reactivity to stress was measured using salivary cortisol samples during the Trier Social Stress Test for Children (TSST-C). Emotional and other linguistic indices derived using Linguistic Inquiry and Word Count software package were meaningfully correlated with children’s depressive symptoms and children’s cortisol reactivity, with medium mean effect sizes of |r| = 0.21 (range: 0.19-0.28) and |r| = 0.23 (range: 0.20-0.29), respectively, across significant correlations, providing preliminary evidence that children’s spoken language coded via automated methods is a valid marker of children’s internal states.

自动编码有可能简化对心理相关的人类行为进行繁琐的手动编码。例如，成人书面语言（例如社交媒体帖子）的自动编码表明，偶然的语言模式与内部心理过程（包括抑郁症状）相关。然而，目前尚不清楚儿童附带口语的自动编码是否与他们的症状和其他风险指数有关。因此，我们检查了 146 名儿童（M_年龄= 9.53 岁，SD  = 0.88；82 名女孩）的口语（转录为自动编码）与其内化症状和皮质醇应激反应之间的关联。儿童完成抑郁和焦虑的自我报告，并在特里尔儿童社会压力测试（TSST-C）期间使用唾液皮质醇样本测量皮质醇对压力的反应性。使用语言查询和字数统计软件包得出的情绪和其他语言指数与儿童抑郁症状和儿童皮质醇反应性显着相关，中等平均效应大小为 | r | = 0.21（范围：0.19-0.28）和 | r | = 0.23（范围：0.20-0.29），分别具有显着相关性，提供了初步证据，表明通过自动化方法编码的儿童口语是儿童内部状态的有效标记。