Comprehension of 500 safety words: a computer-based methodology. 2 volumes.

Abstract

This thesis investigates the relative comprehensibility of 500 words related to consumer safety using a computer-based methodology. Problems in defining and measuring language comprehension are discussed. A pilot study explored techniques for assessing the comprehensibility of 40 safety words and 30 safety phrases. Results determined the feasibility of scaling up the research to a large field study. The main study investigated 500 safety words using questionnaires containing ten each of the target words. Words were randomised and partitioned into administratively convenient sets before generating questionnaires. Questionnaires were tested using national survey techniques. The questionnaire was open-ended and asked respondents for the meaning of each word in the general context of safety. The verbatim results formed the primary data base. Seven "qualified" judges independently categorised the responses according to operationally defined levels of comprehension. The seven sets of judgements were analysed for consensus. Frequency analyses were made on the categorised raw data. The results of several of these analyses were combined to yield tabulations in which words were ordered according to various criteria. Subsequent analysis concentrated on words that looked as if they were either difficult to understand, had provoked significant lack of agreement across the judges, showed a significant number of inappropriate responses or failed to have a large enough sample to yield reliable interpretations. Each word was also analysed by gender and by predicted verbal ability. Finally, conclusions and recommendations are made regarding the methodology and the results it yielded along with shortcomings and potential extensions.