This paper highlights the techniques of natural language processing (NLP) and its applications in survey research, particularly in verbatim text coding and data mining report generation. A prototype system called Codia was developed to incorporate some natural language processing techniques for coding Chinese verbatim answers collected in survey projects. The verbatim answers of different respondents are grouped according to their similarity with a particular answer or input text. As simular answers become neighbours, multiple verbatim answer items can be easily selected and classified. A text generator was developed to generate English description of data mining results. The text generator obtains input of the data mining results and survey questionnaire definitions and outputs a description of the data mining results in English sentences, which are easier to read than the obscure notations obtained from ordinary data mining software.