¹ I have this thesis in a review of Urban's Language and reality and Whorf's Four articles on metalinguistics, Lg. 32.32 (1956): ‘There is, in fact, an amazing suppleness in language for describing the “fullness of reality” and this suppleness is to a large degree due to the shaping and reshaping of vocabulary items to particular contexts. The fineness and adequacy of the items-in-particular-contexts as a descriptive calculus is such that there is probably a sense in which divergencies in vocabulary grids themselves are ultimately irrelevant.‘

² Lew R. Mickelsen, Structural linguistics and machine translation, For Roman Jakobson 345 (The Hague, 1956). The illustrations regarding mutual pinpointing in Russian are found on pp. 447–8.

³ Paul Garvin, Some linguistic problems in machine translation, For Roman Jakobson 183.

⁴ J. R. Firth, Modes of meaning, Essays and studies 125 (London, 1951).

⁵ Zellig Harris, Discourse analysis, Lg. 28.1–30 (1952).

⁶ The phrase is Hockett's: ‘The Pike-trained student will face a practical job: he must . . . above all, translate highly refractory material into the language’ (Two fundamental problems in phonemics, Studies in linguistics 7.29 [1949]).

^6a Noam Chomsky, who kindly read this article before publication, points out that the difficulties here cited can apply to the setting up of equivalence chains as well as concordances. Indiscriminate applications of the method of equivalence chains (as of any of the techniques of descriptive linguistics) can also produce trivial consequences—for instance the reduction of a whole text to a single ‘formula’, regardless of its content structure. It is therefore important to work out a precise account of the factors limiting the application of these methods, as in §1.3 below.

⁷ It follows that a conscientious translator avoids irresponsible and random variations in translating an SL lexical item, but chooses equivalents with due consideration of the requirements of each context. On the other hand, he of course recognizes the impossibility of preserving intact the concordance of the SL text—unless he wants his ‘translation’ to be a grotesque caricature.

⁸ Evaluating translation as such should not be confused with the more recondite question of evaluating language as a descriptive and communicative medium. It seems reasonable to believe that the resources of language as used in free original composition are greater than in its more restricted application in translation efforts. Thus it is sometimes possible to express tolerably well in a somewhat expanded paraphrase the spirit and flavor of a passage, but the translator may be reluctant to expand or dilate one portion of his translation disproportionately lest he distort the document as a whole by unduly laboring one part of it. For some remarks concerning the evaluation of language as a descriptive medium, see my review of Urban and Whorf referred to in fn. 1.