We must now return to the distinction between utterance, sentenceand proposition; three levels of abstraction and idealization which apply to any stretch of language we may wish to translate.
There is a type-token relationship between the three, such that we can envisage the most abstract (the proposition) as being an ideal underlying type of which there are a number of tokens or manifestations: a range of sentences which share the same propositional content. Equally, the same relationship holds between sentence and utterance. Each sentence can be viewed as an ideal type which can be realized by a range of actual utterances; tokens of it.
We are all aware of this distinction between the ideal and the actual in our everyday experience (a point which will be raised in our discussion of the creation of conceptual categories), in which examples abound; the written score and the actual performance of a piece of music; the written text of a play and the production on the night; a recipe and the cooked dish. Music critics, interestingly from our point of view, refer to 'performances' of a piece of music as 'accounts', 'interpretations' and 'realizations', making the same point as we are.
In linguistics, the distinction is crucial and can be exemplified by de Saussure's langue-parole and the similar, though not identical, distinction between competence and performance in Chomsky.
The traditional issues in translation of the relationship between 'fidelity' and 'freedom' and the choice between 'literal' and 'free'(or 'semantic'and 'communicative') seem to resolve themselves into the simple question: 'Are we translating propositions, sentences or utterances?' and, the related question, 'What is the implication of choosing one rather than the other?' This being so, it is essential to be clear in distinguishing the three concepts.
Specifically, the utterance can be typified as being concrete and context-sensitive. It is the utterance and not the sentence that is recorded on paper or an audio tape and it is tied to a specifiable time, place and participants. It is judged in terms of appropriateness rather than grammaticality, i.e. whether and to what extent it is constrained by social convention; whether, in terms of normal expectations of communicative behaviour, it is acceptable.
The sentence in contrast, is abstract and context-free. Unlike utterances, sentences exist (if at all) only in the mind. When a sentence is said or written down, we still tend to refer to it as a sentence. This is an unnecessary confusion. It would be wiser to recognize the difference between the substantial written-down sentence and the abstract idealized sentence of which it is a realization, i.e. the written sentence is better thought of as an utterance or a text. Think of what happens when we remember what someone said or wrote. We tend to remember it in an 'edited'and idealized form; not the actual utterance with its pauses, um's and er's, slips of the tongue, etc., but the idealized sentence of which the utterance we had heard was but one instance. Again, in contrast with the utterance, the sentence is not set in time or space nor tied to any particular participants: speakers, hearers, writers, readers. It is, however, language specific, since it is judged in terms of grammaticality, i.e. whether it conforms to the rules of the particular linguistic code and whether, in those terms, it is possible.
The proposition is even more abstract than the sentence. It is the unit of meaning which constitutes the subject-matter of a sentence (and, once realized in actual use, that of the utterance as well). It has been defined as 'that part of the meaning of the utterance of a declarative sentence which describes some state of affairs' and, hence, in uttering a declarative sentence, a speaker is asserting a proposition.