This paper explores the potential of the use of voice recognition technology with second language speakers of English. The study is a development of an earlier study conducted with a small group of native speakers (, TEXT Technology 8.). The current study involves the analysis of the output produced by a small group of very competent second language subjects reading a text into the voice recognition software Dragon Systems 'Dragon Naturally Speaking'. As the program is speaker-dependent and has to be trained to recognise each person's voice, subjects first spent about 45 minutes reading a training text of some 3800 words. As the test text, they then read a second text consisting of 1050 words. The output produced by the software was analysed in terms of words, sub-clausal units, clauses and t-units. In terms of accuracy, the second language speakers' output on each category of analysis was significantly lower than that achieved by the native speakers. Nonetheless, the results were consistent in line with the native speakers' scores; i.e. that the highest accuracy scores were achieved at the lowest (and most discrete) level of analysis, the word level, and the lowest scores at the t-unit, or sentence level of analysis. The paper concludes that voice recognition technology is still an at early stage of development in terms of accuracy and single-speaker dependency. Nonetheless, the fact that consistent results have emerged suggests that the development of an assessment tool, such as a reading aloud test via voice recognition technology and determining a score through an analysis of the output, may be a testing procedure with potential. Copyright © 1999 Elsevier Science Ltd. All rights reserved.
CitationConiam, D. (1999). Voice recognition software accuracy with second language speakers of English. System, 27(1), 49-64. doi: 10.1016/S0346-251X(98)00049-9
- Voice recognition software
- Reading aloud
- English language