ACES is a both an IM client and an instantiation of a model of linguistic distortions caused by aphasia. ACES distorts messages in a manner similar to those of an individual with aphasia. Our model determines the specific nature and rate of distortions, based upon feedback from study participants and the large body of related literature. In the following sections we describe the underlying model of aphasia distortions and the user interface components of ACES. While ACES was modeled on aphasia, our system could, in principle, be used to emulate other linguistic disorders.
Modeling The Effects of Aphasia
The effects of aphasia can vary widely based on the severity of the impairment, the type of aphasia, and even on the type of word that is subject to error. Therefore, we constructed a modeling system, Aphasia Characteristics Emulation Model Editor (ACE-ME), providing controls that allows the user to configure the degree and type of distortions that will be applied to their messages.
ACE-ME sub-divides the distortions into 5 conceptual categories: Distortion of Content Words, Distortion of Inflections, Distortion of Function Words, Distortion of Fluency, and Other. Each category defines error types that affect similar types of words (e.g., only function words or only content words). By grouping similar errors together, we leverage existing literature that focuses on specific distortion types and provide an intuitive interface for users.
ACE-ME includes the breadth of distortions experienced by individuals with aphasia. Although not every aphasic linguistic error is included, our goal was not to create a linguistically exhaustive emulation tool, but one that was communicatively disruptive in a manner similar to those with aphasia. As knowledge of aphasia and natural language processing expand, so can the capabilities of our system.