Predicting user acceptance of Tamil speech to text by native Tamil Brahmans

This thesis investigates and predicts the user acceptance of a speech to text application in Tamil and takes the view that user acceptance model would need to take into, the cultural constraints that apply in the context and underlines the need for a more explicit recognition. The user acceptance models such as Technology Acceptance Model (TAM) predominantly focus on the technological aspects to determine the acceptance. The cultural variables are considered as external but at the same time they acknowledge the influence of user acceptance due to external variables. The contribution to knowledge is, an empirical link between Tamil usage at a social level that indicates the ability to use and accept Tamil speech to text application. The economic value of Tamil, does not seem to warrant technology use and therefore, speech to text in Tamil was found to be less acceptable in the study samples. In order to achieve the objective of predicting the user acceptance of speech to text in Tamil by the native Tamil speaking Brahmans, the researcher designed and evaluated a paper prototype of an iPhone iOS mobile representation of the paper prototype on the idea of 'what you speak is what you get'. As a result of the researcher’s insider position, the idea was to convert the speech as spoken by the person into Tamil orthography without any technological interference such as auto correct, word prediction and spell check. Due to the syllabic nature of the language and the cultural tendency to code mix and code-switch, the investigation focused on three key areas- code mixing, pronunciation and choice of script . This thesis looks at the complexities involved in accommodating these areas. The user's choice of script was increasingly important as it cannot be assumed that all native Tamil speakers are able to read and write Tamil. In order to bring in rich data, the researcher used the insider and outsider positionality alongside phenomenology. This was also to overcome any potential bias in analysis and interpretation. The multidisciplinary approach to answer the research question was inevitable owing to cultural variables like value and usage of language, social perception of language and its usage specifically code-switching, pronunciation and orthography in the native space. 4 Data gathering was done using quantitative study of transliteration and qualitative interviews of Tamil speaking Brahmans. The findings point to the Vedic philosophical texts and practices that influenced the attitude of the respondents on how words must be pronounced and how they ought to appear in text. The development of the speech to text application could be enriched by using a native approach that embeds cultural and philosophical values. Based on the findings, this thesis has identified areas for further research which is to widely test the user acceptance model proposed in this thesis to aid development of speech to text and to further investigate on native perspective in the wider diaspora and also to investigate cultural and philosophical relevance in speech to text in other languages where technology is in developing stage.