The thesis begins with a brief introduction in which I set out a number of the major questions which I intend to consider. I indicate that the focus for discussion will be functional words. Two questions regarding functional words are asked. First, do they have descriptive content? Second, do they have evaluative content? After a brief discussion, I attempt to set out the significance of these two questions. I point out that both are potentially significant relative to recent controversies over how description is related to evaluation. I indicate, however, that my chief concern will not be with the prescriptivist/naturalist controversy; rather, my chief concern will be with attempting to discover the role played by functional words in both descriptive and evaluative contexts The purpose of the discussion will be to discover whether an analysis of functional words can shed any light on the nature of the fact/value relationship. After raising a number of subsidiary questions, which, it is hoped, will aid the reader in following the argument, I introduce some terminology of which the notion of an F-word is central. A definition of 'F-word' is provided in Chapter II. The 'definition' with which the notion is introduced is simply that an F-word is any word which is like functional words in relevant respects. [page ii of abstract missing] In Chapter IV the discussion turns to a detailed examination of a suggestion of Hare in The Language of Morals. I find it necessary to reconstruct his suggestion and find it contains three elements. Of these three elements I argue that Hare is correct in thinking that in F-inferences, a standard of evaluation is introduced by the use of an F-word. I then argue that Hare is incorrect in thinking that, as in inferences from pure descriptions to hypothetical imperatives, the standard of evaluation in F-inferences is introduced via the conclusion of the inference. Finally, I argue that Hare is incorrect in his view that both hypothetical imperatives and F-conclusions are analytic qua their imperative or evaluative content. I then point out that F-conclusions and hypothetical imperatives; are dissimilcir in this important respect, namely, that unlike hypothetical imperatives, evaluations entailed by F-descriptions are genuine evaluations. Chapter V is a summary of the conclusions which follow from the arguments of the first part of the thesis. I conclude that F-words do have both descriptive and evaluative content. Further, that their descriptive and evaluative content derive from the same source, namely, the fact that F-words identify objects by reference to their function. Because of this, the descriptive content of F-words cannot be separated off from their evaluative content and expressed in descriptive sentences which have no evaluative content. In this sense, F-words and F-descriptions cannot be eliminated. The chapter closes by asking whether there is a second sense in which F-words cannot be eliminated. I ask, 'Are F-words such that to eliminate them from one's descriptive vocabulary is to eliminate the possibility of using descriptive language?' The over all purpose of Part II is to answer the question posed at the conclusion of Part I. I begin that task with a two chapter examination of the relation between perception and goal directed behavior. The basis of the argument is the proposition that if the ability to engage in goal directed behavior is a necessary condition of perception, then the purposes or goals which guide human conduct will be reflected in the ways we identify things. The discussion in Chapter VI and VII revolves around three questions: (i) is perception a necessary feature of goal directed behavior? (ii) is perception itself a form of goal directed behavior? (iii) is there a necessary relation between perception and agency such that it is logically necessary that perceivers are agents? The first two questions are discussed in Chapter VI, the third in Chapter VII. The first question finds a positive answer. A thing which is incapable of perception is incapable, as a consequence, of goal directed behavior. The second question is answered negatively. I point out, however, that perception does exhibit a number of characteristics whose possession suggests that perception has much in common with goal directed behavior. Chapter VII considers the question 'Is there a logically necessary relation between perception and agency such that only agents are capable of perception?'. I argue that: (i) it is logically necessary that something be an agent if it is to be determined that it is a percipient thing; (ii) only if a percipient thing is an agent can it be determined what it is capabl of perceiving; (iii) there is convincing empirical evidence which demonstrates that perceptual skills are acquired in the context of goal directed behavior and further that this does have an important bearing on what someone does in fact perceive; but (iv) the kinds of arguments which attempt to show that percipient things logically must be agents are unconvincing. The next two chapters turn to a discussion of the relation between description and goal directed behavior. I argue that the use of descriptive language for communication is possible only to agents. Descriptive uses of language can be taught only if it is possible to establish publicly what the words in the language mean. A number of arguments are used to show that this is a genuine problem and one which cannot be overcome by non-agents. From this it is seen to follow that for communication to be possible, at least some of the objects about which communication takes place must be identified as objects of manipulation. I argue that it follows from this that for descriptive language users, a fundamental and non-eliminable way of identifying objects is as objects of manipulation. Chapter IX is devoted to an attempt to develop and illustrate the conclusions of Chapter VIII through the use of a model. In the course of the discussion, I show that an object of manipulation is an F-object. Chapter X is the concluding chapter of Part II. I conclude that to communicate using descriptive language f an individual must be able to identify at least some of the F-objects which those with whom he wishes to communicate are able to identify. I suggest that this constitutes a second sense in which F-words are not eliminable. Chapter XI comprises Part III of the thesis. Its purpose is to sum up in a brief way the conclusions of the previous two parts. It also suggests in a highly speculative way some possible implications of the position arrived at in the course of previous argument.