Experimental psychology seems to have been firmly established as a natural science with various experimental devices, mathematical formulae and many other tools. However, we cannot overlook one of the important features of the present-day academic psychology that it seems to be going more and more away from what it was originally intended to be, that is, the investigation into the nature of human mind. In fact, it may be said that experimental psychology does not give satisfactory answers to questions such as those often raised by a layman that are directly concerned with actual functioning his mind. Of course, it would be patently false to argue that psychology has done nothing to meet such a challenge. We have, for example, Freudian psychoanalytic theory. Evidently it has contributed immensely to our understanding of the human mind. A great number of individual experimental studies accumulated in the past fifty years have also done much to advance our knowledge. These contributions notwithstanding, there is a serious lack of a conceptual framework within which results of individual researches can be related to each other properly, or notional and often vague concepts of speculative psychology like psychoanalytic theories are formally described. We have to admit that years of experimental studies in psychology has not resulted in such a framework rich enough in its empirical content to meet reasonable demands on it and to indicate the line of future research. In this paper we have attempted to make a first step toward the construction of such a framework for the problem of memory, which, no doubt constitutes a really fundamental component of psychology. We have first directed our attention to part of the semantics of a group of words in Japanese whose meaning essentially involve the concept of memory. Our analysis makes use of the methods of both. British ordinary language philosophy and the recent theory of generative grammar. Then, we have tried to interpret what the linguistic data seem to suggest in terms of the facts already uncovered through experimental work and the concepts of learning theories. Though it would be far beyond the scope of the present paper to continue this attempt further along this line, we hope that in the course of it a framework will emerge which can incorporate various experimental data on memory and give us a clue to future investigations into its underlying mechanism. Our attempt, then, is in accordance with a methodology that allows not only experimental data but also various ' indirect ' data,, such as those reflected in the conceptual system lying behind our language, to be taken into account. In addition, many intuitive and speculative ideas are fairly freely considered. We believe this does not make our study particularly subjective or non-objective. Various kinds of data are constantly checked against each other and an idea suggested by one of us is examined in so far as it is agreed to by other and is supported by other kinds of more ' objective ' data. This methodology requires much discussion and, to achieve mutual agreement, the members who joined a study by this methodology must have some common background. It can only be applied when they have opportunities to have a great deal of discussion in which speculations and intuitive judgements must be exchanged. Thus a science of human mind by this methodology may be called a round-table psycholygy, as the tile of this paper represents.