Programming by voice : Efficiency in the Reactive and Imperative Paradigm

Abstract: Due to the recent increase of computer programmers, and asa result, sufferers from Repetitive Strain Injury (RSI),programming by voice is becoming an important alternativeto keyboard-based input. Using speech recognition softwareis more difficult for programming tasks than naturallanguage. This is because writing code requires punctuation,symbols, and abbreviations that need to be spelled out. It isalso not limited to writing code, but also includes navigation,editing, and error correction. Therefore, an easy and flexibleway of working with code by voice is therefore. This studyexamines the efficiency of vocal programming in differentparadigms (namely, the imperative and reactive paradigm).We survey which advantages and challenges the reactiveand imperative paradigm have within this field. Bydiscussing if and how the reactive paradigm could beincorporated into future tools and covering the mostcommon challenges, we hope to provide a base for furtherstudies on programming reactively by voice. A brief surveyof tools for vocal programming is carried out. Using TalonVoice, examples written in Java and RxJava, we measure theperformance of the respective paradigms, by recording thenumber of words, syllables, and necessary pauses used whenproducing code vocally. The results vary in vocal load, butboth paradigms have advantages in vocal programming. Theresults also show that reactive programming, thoughgenerally requiring more code for the same tasks asimperative code, does not require the vocal programmer toutter as many words per character as within the imperativeparadigm. The authors discuss the effects of cognitive load inrelation to the presented findings.