Information plays various roles for supporting human activity. Its basic role is to describe objects in the world. Form of representation is first decided, then method of its processing can be defined on its basis. Thus an information processing paradigm is defined. As human activity gets more and more complicated, information also becomes more and more sophisticated. Form of representation and processing method become the more complicated. Sometime in human history symbolic language has been developed. It facilitated representation of things. With the progress of language, information became the more sophisticated one. Development of variety of paradigms became possible by means of language. Each of which has a specific scope of applications There is a lot of information, however, that cannot be captured by symbolic language. Representation and processing of information unable to be symbolized, that is, non-symbolic information processing paradigm is very different from any of symbolic information processing paradigms. These two groups of paradigms divide the whole information world. Most applications have been done within either one of them so far. Progress of human activity requires a paradigm with a larger scope. If any of existing single paradigms cannot satisfy this requirement, a new paradigm must be developed. Integration of different paradigms is becoming a very important approach. However conversion between paradigms is necessary for the purpose and conversion between substantially different paradigms, such as between symbol processing and non-symbol processing,is very difficult. Bridging the gap between these two paradigms is becoming an important issue and is attracting much interest of researchers. Discovery and data mining is one of such issues. There is an interest in the more basic question such as how could human being acquire language? To discuss these issues is the major objective of this talk.