- Every day John visits his sister Ann to see his nephew Sam.
- Every day John visits his sister Ann to see her nephew Sam.
- Every day John visits his sister Ann to see her grandchild Sam.
- Every day John visits his sister Ann to see his grandchild Sam.
The last translation seems less probable than the previous ones, but it is possible to imagine a world in which (i) John had a son, Max, (ii) Max married Mary, (iii) Max and Mary gave birth to Sam, (iv) they then died in a car accident, (v) aunt Ann decided to adopt Sam. A very improbable but not impossible world situation (given the present corruption of any moral principle) would be that Max, son of John, had a love affair with his aunt Mary, and that they produced Sam, so that Sam can be correctly defined both as John’s grandchild and as John’s nephew. In any case, if the general context does not provide translators with further information, so that they can conceive of a possible world structure, it is impossible to translate that Italian sentence on mere linguistic grounds.
In order to understand a text, or at least in order to decide how it should be translated, translators have to figure out the possible world pictured by that text. Often they can only make a hypothesis about that possible world. This means that a translation is also the result of a conjecture or of a series of conjectures. Once the most reasonable conjecture has been made, the translators should make their linguistic decisions accordingly. Thus given the whole spectrum of the content displayed by the dictionary entry (plus all the necessary encyclopaedic information), translators must choose the most suitable or relevant meaning or sense for that context.
You have probably realised that I started speaking of differences between languages (in the sense of tongues) and now I am speaking of differences between texts. This is a very crucial point for every translation theory. But let me go on with languages.
These are two main arguments against translation, both more germane than they seem at first sight: (i) the impossibility of setting up a unique translation manual and (ii) the incommensurability of language structures.
As for the first argument, following Quine’s famous example, it is difficult to establish the meaning of a term in an unknown language even when the two speakers are facing the same external stimulus – which casts in doubt the notion of referential equivalence. If the linguist points to a passing rabbit and the native informant utters gavagai! it is uncertain whether, by that expression, the native speaker means rabbit, stages of rabbit, or the fact that an animal is passing through the grass at that moment – or whether gavagai is the proper name of a universal rabbithood.
It is impossible to decide, if one has no previous information on the native culture – that is, if one does not know how the natives categorise things, parts of things, or events involving things. The linguist must start by making a series of analytical hypotheses in order finally to produce a translation manual that corresponds to a whole anthropological handbook. However, the indeterminacy of translation is proven by the fact that rival systems of analytical hypotheses can produce different (but equally legitimate) translations of the same sentence. In Quine’s words, ‘Just as we meaningfully speak of the truth of a sentence only within the terms of some theory or conceptual scheme . . . so on the whole we may meaningfully speak of interlinguistic synonymy only within the terms of some particular system of analytical hypotheses.’4
In spite of the current cliché about the incompatibility between Anglo-Saxon and Continental philosophy, I believe that Quinean holism is not so different from the idea that every language has its own genius (as Humboldt said) or – better – that every language expresses a different world view (the Sapir–Whorf hypothesis). In what sense a given language designs its own world view is pretty clearly explained by the structural semiotics of Hjelmslev.5 According to Hjelmslev, a natural language (and, more generally, any semiotic system) consists of a plane of expression and a plane of content which represents the universe of the concepts that can be expressed by that language. Each of these planes consists of form and substance and both are the result of the organisation of a pre-linguistic continuum.
If we consider the English language, the form of expression consists of its phonological system, its lexical repertory and its syntactic rules. Through this form of expression, we may generate various substances of expression, such as the words we utter every day. But I shall speak of substances in another essay. Let me consider here the problem of form. In order to elaborate a form of expression, English has selected (from the continuum of the sounds that the human voice can produce) a series of sounds, excluding others that exist and are pronounceable but which do not belong to the English language.
If the sounds of a language are to be comprehensible it is necessary to associate them with contents. For Hjelmslev, the continuum of content is the totality of all that can be thought and said: namely the whole universe, physical and mental (as far as we can speak of it), including the material elements such as sounds that we use to make expressions. Each language organises the universe of what may be said and thought into a form. The system of colours, the organisation of the zoological universe in genera, families and species, the opposition of high vs low, and even the structures that linguists identify in phonological systems – all belong to the form of content. Certain cultures organise the kinship system by isolating differences that in Great Britain, as well as in Italy, are ignored: for instance, the difference between the brother of the mother and the brother of the father, who for us are both uncles.
In this sense two semantic systems can result in being mutually inaccessible, because they segment the content continuum in a different way. Thus, according to Quine, one cannot translate the expression neutrinos lack mass into a jungle language, and one only need think how difficult it is to translate the German word Sehnsucht into English or Italian in order to understand that German culture has the precise notion of a certain passion whose ‘semantic space’ can be only partially covered by terms like nostalgia, yearning, craving for or wishfulness (and none of them renders it adequately).
But frequently such inadequacy is only the result of a sort of ‘optical effect’. Let us return for a moment to the distinction between nephew/niece and grandson. It is clear that where English recognises three separate content entities, Italian recognises only one: nipote.
ENGLISH ITALIAN
Nephew
Niece Nipote
Grandchild
Now, it is true that in Italian a single word is used to express three different units deriving from the segmentation of the continuum of kinship relationships. But this graphic representation of the phenomenon can be very misleading if we imagine that the four words (nephew, niece, grandchild and nipote) represent concepts, or content units. They do represent words, which are used to mean a certain content unit, and curiously Italian has one word for three distinct positions in the kinship system. But the fact that there is only one word does not mean that Italians do not see any difference between the child of one’s son or daughter and the child of one’s sister or brother. They see it to such an extent that even in the case of death duties the two kinds of relatives pay a different tax. This means that Italians have more linguistic difficulties than speakers of other languages in certain contexts where the distinction between nephew/niece and grandchild becomes relevant.
Despite having only one word, then, Italians conceive three content units, and these units, which are not distinguished from each other by a specific word, are nonetheless interpreted by dictionaries or encyclopaedias, which explain what I have just explained – namely, the difference in kinship between a niece and a grandchild. To translate means to see, under the words, the possible interpretation of these slots, which are not conceptually empty but only linguistically ill-named.
ENGLISH TERMS CONTENT ITALIAN TERMS
Nephew Son of the brother or sister
Niece Daughter of the brother or sister Nipote
Grandchild Son or daughter of the son or daughter
To be honest, since there are different kinship systems, more finely segmented, one could say that even English is very savage in comparison with a jungle language that has more names for different kinship positions. For example:
ENGLISH TERMS CONTENT TERMS OF JUNGLE LANGUAGE X
Nephew Son of the brother Term A
Son of the sister Term B
Niece Daughter of the brother Term C
Daughter of the sister Term D
Granchild Son or daughter of the son Term E
Son or daughter of the daughter Term F
But this does not prove that an English speaker is unable to recognise the different kinship positions expressed in our jungle language by the series of terms A–F. English speakers have no names for these different positions but are able to interpret each of them correctly.
The challenge for a translator, when two languages seem to have a different segmentation of the content continuum, is to make a reasonable conjecture about
