Stuttering in the real world: Quantifying variability to improve measurement reliability and validity

项目摘要(30行) 口吃是一种高度可变的情况：人们在不同的时期表现出不同程度的口吃行为 任务，在不同的场合、不同的时间与不同的人交谈。尽管可变性是 被广泛认可的，大多数对口吃行为的评估，无论是在临床环境中还是在 对于研究，使用在少量环境中收集的简短语音样本，这些样本不能反映 自然交流(例如，独白或朗读)。不幸的是，我们不知道这些 临床/实验室措施代表了人们在现实世界中的说话经验。样例 有证据表明，它们不是，但绝大多数关于口吃的临床工作和研究仍然 没有考虑到这种可变性。这引发了人们对现有研究的有效性的质疑，即 依赖于可能不能反映人们真实口吃行为的衡量标准。 在这个项目中，我们试图克服这些关于口吃测量有效性的不确定性。 我们将收集有史以来最大的口吃患者的日常语言样本 连续记录7天内的对话互动。这将使我们能够 检查人们在不同情况下，随着时间的推移表现出的各种口吃行为。 对于目标1，我们将使用这些史无前例的大量语音样本来测试当前的做法是否 在临床或实验室环境中收集简短的语音样本就足够了。试点数据和先前的研究 建议采取更广泛的措施，以便更好地反映人们的真实口吃 行为。对于目标2，我们将使用潜变量建模来开发口吃的测量模型 以支持评估口吃的新方法，这些方法更能反映现实世界的可变性。为了达到目标 3，我们将使用结构回归模型来探索关键个人特征、 例如一个人对口吃的反应方式，有助于体验变化无常。为 目的4，我们将进行一项混合方法研究，以探索变异性如何影响 口吃对人们生活的影响，最终目的是改善对口吃的评估，因此 我们可以减少不利影响，提高口吃者的生活质量。 总体影响：我们的发现将增加我们对导致变异性的因素的理解 并为改进临床和实验室中的口吃评估提供必要的指导 设置。我们将通过以下途径提供我们精心转录、带注释的演讲样本 这样其他研究人员就可以使用我们的数据来回答许多关于 在现实世界中结巴。我们的工作有可能打乱目前人们对口吃的看法 同时提供有关研究人员和临床医生如何 更好地评估口吃，以收集人们在日常生活中流畅程度的更具代表性的衡量标准。

Project Summary (30 lines) Stuttering is a highly variable condition: People exhibit different amounts of stuttering behavior in different tasks, when talking to different people, in different settings, and at different times. Although variability is widely acknowledged, most assessments of stuttering behavior, whether conducted in a clinical setting or for research, involve brief speech samples collected in a small number of settings that do not reflect natural communication (e.g., monologuing or reading aloud). Unfortunately, we do not know if these clinic/laboratory measures are representative of people's speaking experiences in the real world. Ample evidence suggests that they are not, but the vast majority of clinical work and research on stuttering still does not account for this variability. This raises questions about the validity of existing research that relies on measures that may not be reflective of people's true stuttering behavior. In this project, we seek to overcome these uncertainties about the validity of stuttering measurement. We will collect the largest-ever samples of the everyday speech of people who stutter through continuous recording of conversational interactions over a 7-day period. This will allow us to examine the full range of stuttering behaviors that people exhibit across situations and over time. For Aim 1, we will use these unprecedentedly large speech samples to test whether current practices of collecting brief speech samples in clinical or laboratory settings are adequate. Pilot data and prior studies suggests that a broader range of measures will be needed in order to better reflect people's true stuttering behaviors. For Aim 2, we will develop a measurement model for stuttering using latent variable modeling to support new methods for evaluating stuttering that are more reflective of real-world variability. For Aim 3, we will use structural regression modeling to explore the ways in which key personal characteristics, such as the ways in which a person responds to stuttering, contribute to the experience of variability. For Aim 4, we will conduct a mixed-methods study to explore how variability affects the adverse impact that stuttering has on people's lives, with the ultimate goal of improving the assessment of stuttering so that we can reduce adverse impact and enhance quality of life for people who stutter. Overall Impact: Our findings will increase our understanding of the factors that contribute to the variability of stuttering and provide needed guidance for improving assessment of stuttering in clinical and laboratory settings. We will make our carefully transcribed, annotated speech samples available through FluencyBank, so that other researchers can use our data to answer many unanswered questions about stuttering in the real world. Our work has the potential to disrupt current thinking about stuttering measurement while simultaneously providing a solution about how researchers and clinicians can better assess stuttering to gather more representative measures of people's fluency in their daily lives.