Task Dynamics in the Production of American Sign Language

描述（由申请人提供）：本研究旨在开发美国手语（ASL）中第一个手势运动的任务动态模型。该模型将从运动学标志生产数据的分析中得出。我们的方法是基于发音音韵学理论，该理论提出语音的基本单位是发音手势。这些手势是运动任务，从一个产品到另一个产品是不变的，它们的时间重叠决定了语言输出的结构。在这项研究中，我们假设发音手势也是符号的结构基本要素，我们试图发现手势是什么以及它们是如何计时的。在符号制作过程中，发音器必须从一个符号的位置移动到下一个符号的位置，并且它们还必须产生个体符号内部的运动。符号音韵学理论试图区分这两种类型的运动（符号内部和过渡），并确定运动是符号的特定部分，还是仅仅是从一个物理位置到另一个物理位置的一种手段。本研究将探讨美国手语中手语内部和过渡运动的运动学和动力学。运动学数据将从本地手语者收集，因为他们产生的手势运动接近或远离身体上的特定位置。为了确定当运动轨迹和速度明显变化时，手势产生的哪些方面保持不变，将操纵签名速度、符号上下文和短语上下文。来自运动学数据的测量将被用作计算模型的输入，这将允许我们检查发音手势。从计算模型中识别出的手势将通过动画程序与人类生产数据进行比较，并由本地ASL手语使用者判断其可理解性和自然性。通过这种方式，我们将使用收敛的证据线来测试符号生产的最小单元是什么，以及它们是如何相互关联的。这项研究与美国国立卫生研究院的公共卫生使命有关，因为它有助于我们了解美国手语的结构，北美估计有50万人使用手语。这项研究有可能提高聋人手语使用者的生活质量。我们的方法可以通过自动化通信系统为聋人提供更自然的交流方式。如果我们提出的结构准确地描述了符号结构的最小单元，那么它应该为符号合成提供最自然的模型。

DESCRIPTION (provided by applicant): The proposed study seeks to develop the first task-dynamic model of sign movement in American Sign Language (ASL). The model will be derived from analyses of kinematic sign production data. Our approach is grounded in the theory of Articulatory Phonology, which proposes that the basic units of speech are articulatory gestures. These gestures are movement tasks that are invariant from one production to another, and their temporal overlap determines the structure of language output. In this study, we hypothesize that articulatory gestures are also the structural primitives of sign, and we seek to discover what the gestures are and how they are timed. During sign production, the articulators must move from one sign's location to the next sign's location, and they must also produce movements that are internal to an individual sign. Theories of sign phonology have tried to differentiate these two types of movement (sign-internal and transitional) from one another and also to determine whether movement is a specified part of the sign, or simply a means of getting from one physical location to another. This study will examine the kinematics and the dynamics of sign-internal and transitional movements in ASL. Kinematic data will be collected from native signers as they produce signs with movements toward or away from specific locations on the body. Signing speed, sign context, and phrase context will be manipulated in order to identify what aspects of sign production remain invariant when movement trajectories and velocities are explicitly varied. Measures derived from the kinematic data will be used as input to a computational model, which will allow us to examine the articulatory gestures. The gestures that are identified from the computational model will then be compared to human production data via an animation program and judged for intelligibility and naturalness by native ASL signers. In this way, we will use converging lines of evidence to test what the minimal units of sign production are and how they are related to each other. PUBLIC HEALTH RELEVANCE This study is relevant to NIH's public health mission, because it contributes to our understanding of the structure of American Sign Language, which is used by an estimated 500,000 people in North America. The study has the potential to enhance the quality of life for Deaf sign language users. Our approach could lead the way toward a more naturalistic means of communication for Deaf people via automated communication systems. If the structure we have proposed accurately describes the minimal units of sign structure, then it should provide the most natural model for sign synthesis.