One of the wonders of language is that it can be used to talk about itself. The language thus used is called meta-language and the language talked about is called object language. An object language may not be combined with the meta-language, such as in an article in Chinese on English, where English is the object language while Chinese is the meta-language. If the article is in English, then both the object language and the meta-language are English but still the two are separated in terms of space. Sometimes, an object language is combined with the meta-language, as in: 

The statement separately printed here is false.

 Obviously, the above statement says something about itself and therefore it is both meta-language and object language.

 This article discusses the translation of meta-language. Before that, I would like to say a few words about the truth value of a translation.

 In logic we distinguish between a sentence, a proposition and a statement. A sentence is a string of words put together according to a certain grammatical rule. "Beijing is the capital of China" and "Ah, Yellow River, you are great , like a giant standing on the Asian plain" are both sentences. A proposition is expressed by a sentence and has truth value, while not all sentences have truth values. In the second sentence here, there is no truth value but a kind of feeling. A statement is more an action than a product of articulation, namely what one says about something. In the sentence "The sun is in the Milk Way Galaxy, there is a statement about the sun. so we can say that a statement is an action, a sentence is an entity and a proposition is a kind of meaning. We express proposition with sentences to make statement about things.

 Only propositions have truth value. When we say a sentence is true, we are actually talking about the proposition expressed therein.

 Translation should be faithful to the original. This of course entails that a translation should preserve the truth value of the original sentence. In most cases translation is truth preserving. Not only is this truth in the translation of ordinary sentences, but also in the translation of meta sentences. For example the above sentence can be readily translated into China without changing the truth its expresses.

 But sometimes a translation may change the truth value of a meta sentence. For example the sentence "The sentence you are reading now is written in English" is true only when it remains in English. If we translate it into an other language, it will not be true anymore. So we can see that some meta sentences are language-bound.

 From this example we can conclude that translatability has something to do with the meta function of language. In the rest of this article I shall discuss whey some expressions can be translated and why some cannot. When talking about the translatability of meta sentences, I am actually talking about if the object sentence talked by the meta sentence can be translated or not. I shall point out that some difficulties in translation are in fact the manifestations of meta-language.

 In the field of translation theory, those who noticed the relation between translatability and meta-language include Peter Newmark 91981), who discusses the translation of source language (henceforth SL) texts that bear strong linguistic features of the SL and the translation of literary games in non-literary works, such as puns, proverbs, words of multiple meanings, metaphors, etc. What he says about the translation methods of these language games is not new. What is new is that he approaches the issue from the point of meta-language. But he does not point out clearly why these should be regarded as meta-language and why they are difficult to translate and even untranslatable. We can also ask a further question: are they all difficult to translate and untranslatable?

 In book on the meta function of language, the definitions and examples are both confined to the meta-language that is used to talk about the object language, for example:

  Meta-language: a language that is used to describe the structure of another language; any language whose symbols refer to the symbols of another language (Mario Pei and Frank Gaynor, 1954, p135, here the English is a back translation from the Chinese)

  Meta-language, rational language: the language or a set of symbols that is used to analyze and describe another language (the observed language or object languages), such as a word used to explain another word or the native language used in foreign language teaching (Hatman and Stock, 1981, p 213. The English here is a back translation from the Chinese).

Another often mentioned meta sentence is that which is both meta-language and object language, as the example I give earlier. Some figures of speech are treated as meta-languages mainly by Newmark and Jacobson, who did not argue for the necessity of such an approach.

When meta-language and object language are in one sentence, it is sometimes difficult to separate them. In the following dialogue between mother (M) and daughter (D) on grammar, there is such a question (Dialogue taken from Bolinger and Sears, 1981, p175):

M: How about this one: Boy is at the door.

D: If his name is Boy. You should---the kid is named John, see? John is at the door or A boy is at the door or He's knocking at the door.

In M's question, is fairly easy to identify meta-language and object language, but in D's explanation they get mixed up. If we translate the dialogue into Chinese literally the translation will not be understood. We have to keep the object sentences here in the original form and add some words to make it clear to the reader that they are talking about how to say some English sentences.

I would suggest we classify meta-language sentences according to the extent to which meta-language can be separated from the object language into three kinds: Conspicuous Meta Sentence (CMS), Semi-Conspicuous Meta Sentence (SCMS) and Obscure Meta Sentence (OMC).

In CMS, meta function and object function are separated in space. In most cases, there are always some marks indicating what is the object language, such as " so called, so to speak, sometimes known as, etc." (Newmark, 1981, p110) or we can used expressions such as "the sentence, the words":

1) The sentence "Bill hits John" is an SVO sentence.

2) The words azure, pleasure, measure, Asian, fusion, and a good many more have the

phoneme /z/ in the middle of a word.

In SCMS, meta-language and object language are either separated in time but not in space or in neither case:

3) a. The following sentence is false.

b. The above sentence is true. 

Here (a) and (b) are both meta-language and object language, but (a) becomes an object language only after (b) is produced, which means that the meta function and the object function of (a) are separated in time while (b) is immediately a meta-language and an object language as soon as it is produced. 

OMC is hidden in some expressions that do not have the form of a meta sentence but depends on the thuthfulness of a meta sentence for their effectiveness. Plays on words belong to this category. For example in "A river has two banks, therefore it is rich", the word "bank" is a pun. In this sentence there implied a meta sentence about the word: "The word ‘bank’ has two meanings". This meta sentence is a true statement.

 SCMS can be turned into CMS. For example the sentence "the sentence you are reading now is false" can be turned into "The sentence ‘the sentence you are reading now is false’ is false." As to OMC, we can extract a meta sentence from it.

 The nature of all kinds of word plays is the same, namely to play on words, which requires that there must be some statement about the words. Therefore, word plays fall into the category of metal-language because they imply a meta sentence.

 Some word plays are purely meta functional, such as the example about the word "bank", which says nothing except stating something about the word. But not all word plays are so. Many conveys some cognitive meaning at the same time. Poetry, despite its meta-language function in its rhymes and rhythms, is not about language only but also conveys feelings and even meaningful statements about the world.

 We can classify word plays into three kinds according to the cognitive content they carry. One is the kind that conveys true cognitive knowledge, another is the kind that conveys pseudo-cognitive knowledge and the last one kind does not convey any cognitive content. The first includes poems and some figures of speech in articles. The second includes such examples as the riddle about the river. To say that a river is rich is obviously absurd. The third kind includes pure word plays:

  "Twas brillig and the slithy toves

Did gyre and gimble in the wabe

All mimsy were the borogroves

And the more raths outgrabe.

 Then, into what category shall I put the word play that I am going to discuss in this article? Since I am talking about translation, I have to ask a further question: Are these word plays language bound or not?

 Obviously many word plays are language bound. Puns, for example, are a typical case. Puns are true only with regard to the language they belong to. In the title "Fifty Years of Scarlet Fever" (Time, June. 25, 1988), "Scarlet Fever" refers to the popularity of the novel Gone with the Wind and at the same time has some humorous effects because it also is the name of a disease. The meta statement here is "The word ‘Scarlet Fever’ has two meanings in this title." This statement is true only when the title is not translated, say, into Chinese.

 Some puns depend on cultural background. Historian Susman wrote such a paragraph in his Culture as History:

  In the beginning there are the words, and all kinds of words from all kinds of places...Thus the historian’s world is always a world of words: they become his primary data, from them he fashions facts.

 Here the author cleverly quotes a sentence in John’s Gospel with slight change: "In the beginning there is the word". This play on the sentence implies a meta statement about the sentence that is only true in the English culture. In Chinese it would not be true, because the standard Chinese translation of the first sentence in the gospel remains standard and authoritative and cannot be changed readily: "Taichu youdao"( In the beginning there is the Tao).

 Some meta-language is not language bound. An advertisement in Life goes like this: "Life would be empty without Kodak professional film." In this sentence "Life" and "empty" both have double meaning. ‘Life" refers both to the magazine and to real life and "empty" refers both to a blank space in the magazine and to the meaningless of life. In Chinese Life is translated into "shenghuo", a word meaning "life". So the pun here can readily be translated into Chinese with the same effect.

 Some meta-language seems to be language bound, but actually not. The meaningless poem we cite above has been translated into Chinese by the famous linguist Yuanren Chao. By using meaningless words in Chinese, the rhetorical function of the original is well preserved.

 Most words in language have reference. Only so can they be used to talk about the world. This is true for the words used in meta sentences too. Because mete sentences are used also to talk about language itself, the words in them have the function of self-reference. In language communication, the message conveyed by the self reference function and the reference function of meta sentences varies with communication situations. I have said that word plays that are language bound are not translatable. But this needs qualification. Whether they are translatable depends on the weights of self reference and reference in the conveyed message. The following are some more examples to illustrate my point:

  1. The first word of this sentence has three letters.

 Viewed in isolation, this sentence cannot be translated into Chinese, since after translation we get a false statement about the first word (it will be the first character in the Chinese version). But in communication, this sentence is an example of meta-language and we can create a Chinese sentence that serves the same function. The Chinese version will then read as "The first character of this sentence has five strokes."

  2. In an English humor, when a US battleship visited England, the signal reads "Next to None". When it entered the port, however, it found moored beside it a small ship with the signal reading "None".

 In this humor, the self reference and reference have the equal weight in the message and thus it cannot be translated into Chinese.

  3. Story; history; His story! This is the heart, the unity, the crux of the bible. We may well call it the crux, for the Cross is at the center of the story. (Donald Cogan: The English Bible)

There are two word plays in this example. One is to turn the word "history" into "His story" and another is the pair of words that sound similar: "crux" and "Cross". The first word play serves to better convey the meaning of the book and therefore its self reference is less important than their reference. In translating, we just translate the reference.

In the second one, the self reference has more weight than reference. If we translate into Chinese the reference only, the translation is hard to understand. The translation would read like "We may well call it the essence, because the Cross is the center of the story." In translation, the relation between "essence" and "Cross" cannot be made clear. A better one will translate "crux" as "shizhi"(actual nature) and "Cross" as "shizijia" (a frame that looks like the Chinese character for "ten"), but the graphic effect is lost here.

From the above analysis we can see that viewed in isolation, word plays depend on whether they are language bound for their translatability. All the word plays that incorporate a meta sentence that is true only in the original language will not be translatable. Viewed dynamically and from the point of view of communication, the translatability of a word play depends on the weight of the self reference and reference of the implied meta sentence in the message of communication.

References

Brower, R. ed. On Translation (Cambridge: The University Press, 1959)

Bolinger, D. and D. A. Sears Aspects of Language 3rd ed (New York: Harcourt Brace Ivanovich, Inc. 1981)

Jakobson, R. "On Linguistic Aspects of Translation" in Brower ed. 1959

Newmark, P. Approaches to Translation (Oxford: Pergaman Press, 1981)

Pei Mario and Frank Gaynor A Dictionary of Lingusitics (New York: Philosophical Library, 1954)

 References in Chinese omitted

 This is a translation of an article I published in the Issue III of Jiaoxue Yanjiu in 1988.