On translatability from the perspective of Wittgensteinian prototype: translations of “Dao” as exemplars

1. Introduction

Debates on translatability, according to Steiner (1998, 77), may be grouped under two headings: the universalist view and the monadist view. In the universalist view, common human rationality and experience determine the possibility of conveying all possible meanings and thus translation is inherently possible. As was observed by Jakobson, languages “differ essentially in what they must convey and not in what they may convey” (1959/1971, 264) (emphasis in original). The monadist view doubts whether fully adequate translation is possible. According to the monadist view, each language embodies different conceptualizations of the world through its unique lexis and grammar. The conceptualizations build different life worlds and thus make translation impossible.

Translatability, whatever view people hold on it, can be approached descriptively and/or prescriptively, with positivist and/or post-positivist orientations. Fundamental as these orientations are, we can follow Wittgenstein’s thinking to approach the issue of translatability from a new perspective. As is shown by Wilson (2015), using the later philosophy of Wittgenstein is of use to translation. In this article, we are going to discuss whether translatability is possible and how we can judge translations from the perspective of Wittgensteinian prototype, with translations of “Dao” in Daodejing as instances.

Daodejing (also known as Laotzu or Tao Te Ching) is a Chinese classic, which is believed to be written around the sixth century BC by Laozi (traditionally spelled as Lao Tzu in English). The text of Daodejing is fundamental to Taoism, either in philosophy, or in religion, or in everyday life. “The proof of classical status, the timeless character of a work, is attained by means of survival in time” (Lianeri and Zajko 2008, 1). Daodejing is such a classic which survived more than two thousand years. To non-Chinese readers/speakers, it is full of attractions. As Venuti observes, “the foreign-language texts we call ‘classics’ do not merely attract translation, but eventually, when their copyright expires, become subject to multiple retranslations” (2008, 27). This is particularly true for the case of Daodejing. Since it began to appear, it has been interpreted in various versions. In the recent two hundred years, Daodejing has been translated and retranslated into western languages more times than any other Chinese works, either by Chinese or non-Chinese translators. Among the translators, some favor rendering of the original text as faithfully as possible while some others prefer interpretations for varied purposes. For example, Eoyang (1993, 71) criticized such translations as Witter Bynner’s for deviating from the original text while LaFargue (1992) and Herman (1996) argue that translations with poor scholarship have their role in meeting the spiritual need in the west.

To render or interpret the original text of Daodejing, surely, is not easy at all. Being terse, poetic, deliberately ambiguous, the text of Daodejing encourages interpretations of any kind. What is worse, there are no punctuation marks in the original, and to move a full stop or a comma a few words forward or backward can profoundly alter the meaning of the original. Such alterations are reasonable in certain contexts and they result in quite different translations. Problems as such in translating Daodejing can be best illustrated by the core term of the source text: Dao. So far as we can see in 175 English translations of Daodejing (see BPS 2014), 106 versions use “Dao” or “Tao,” 41 versions choose “Way,” and 28 versions use other choices. These “other choices” range from “God” by G.G. Alexander (1895; see BPS 2014), “The Providence” by E.H. Parker (1903; see BPS 2014), “The doing” by Tom MacInnes (1927; see BPS 2014), and “Existence” by Witter Bynner (1944; see BPS 2014), to “Way-making” by Roger T. Ames and David L. Hall (2003; see BPS 2014) and “The Cosmic Consciousness” by Brian Donohue (2005; see BPS 2014). The abovementioned translations do reflect part of the core term, but none of them is adequate in representing the complete meaning of “Dao.” As transliterations, “Dao” and “Tao” seem to reflect completely the meaning of the term, but they are at present inaccessible to numerous English readers, far from being counted as translation.

Therefore, the questions we may consider here are: Is translatability possible? If not, why do we have so many translations of such classics as Daodejing and many other translations are still on the way to come to literature? With so many translations of a text, how do we go on to judge what might be a “good” translation? In order to answer these questions, we will first outline Wittgenstein’s notion of prototype, and then observe translatability from the perspective of Wittgensteinian prototype. “Dao” and its translations will be used as instances.

2. Wittgenstein’s notion of prototype

Prototype theory proposed by Rosch has been extensively used in many disciplines, and those applications have already yielded numerous findings, among which some are concerned with translation. In practice, for example, Sobashima, Furuse, and Iida (1994) proposed a bidirectional translation prototype through calculating the similarity scores and selecting the most appropriate expressions; Penn and Carpenter (1999) discussed how Attribute-Logic Engine (ALE) engine can be a prototypical translation system; Pastor et al. (2001) suggested that EUTRANS can be a speech-to-speech translator prototype; Wray et al. (2004) focused on TESSA, an interactive translation system between deaf clients and postmasters, and held that the system is a live translation prototype.

In theory, “prototypology” was proposed in the 1980s and text-type was classified according to prototypical analysis (Neubert and Shreve 1992). “Translation” was then considered a prototypical category which embodies the prototype effect, i.e., graded difference between members of the category (Halverson 2000). Nida (2001, 34) mentioned “typicality” and “prototype” on the lexical level but no further discussion was found in his writings. It is Snell-Hornby (1988/1995, 3) who suggested that the gestalt approach should be advocated in translation studies. She then proposed three strategies in translation: (1) From the SL variant the translator must recognize the SL original (or prototype); (2) the translator must know (or find out) the corresponding TL prototype; and (3) the translator must adapt the TL prototype to recreate the SL variant (98). These applications are constructive and the thinking on translation via “prototype” is illuminative, but “prototype” in these writings is different from that in the Wittgensteinian sense. In general, prototype is considered similar to the model of something, the object or the original (Hermans 2009, 179), and as a typical example, pseudotranslations have often served as prototypes (i.e., the starting point) for new literary genres (Rambelli 2009, 210).

To put it in a few words, “prototype” in the literature of translation studies, both practically and theoretically, refers to something concrete and original which can be the starting point or the background, a model on which later stages are judged or based. It is known that “prototype” is usually defined as: (a) the model or proto-image of all representatives of the meaning of a word or of a “category,” for example, Shakespeare can be regarded as the prototype of the category writer; (b) the bundle of typical features of a category; and (c) the features themselves can also be more or less typical (Bussmann 2006, 963–4). These explanations in fact originated mainly from studies carried out by Rosch (1973; 1975; 1978) and her colleague (e.g., Rosch and Mervis 1975), while Rosch’s research was basically built on Wittgenstein’s thinking (see Rosch [1999] for a summary of her studies).

Scholars who follow Rosch then equate “prototype” with “best exemplar,” for Rosch in her early writings defined prototype as “the clearest examples” or “the best examples” (Rosch 1975, 544). But as Rosch (1999) later summarized, prototype is more than “the best exemplar,” and “family resemblance” by Wittgenstein should be the essence of prototype, but in Rosch’s experimental studies, something should be considered as a “prototype” of a category for the feasibility of carrying out experiments. In other words, there is no such a thing as “prototype” at all in categorization. Rosch (1999, 65) refined her definition and proposed to take “prototype” as “the best judged exemplar” (my own emphasis). The word “judged” here is crucial, for it means this so-called “best exemplar” is true only to experimental conditions. In real life, it is only true to a very limited context. Unfortunately, Rosch’s early definitions have already misled so many people in understanding “prototype” in the Wittgensteinian sense.

As we can find from Wittgenstein’s writings, the term “prototype” should be noted for the following aspects:

1. Prototype is distinct from any particular exemplar. In Notebook 1914–1916, Wittgenstein (1961/1998, 33) asked if one can negate a picture. Wittgenstein wrote, “I can only deny that the picture is right, but the picture I cannot deny” (33) (emphasis in original). In other words, we can negate a proposition, but a picture cannot be negated, which suggests that propositions do not necessarily represent reality in the world. Similarly, if prototype is uniquely concerned with naming, it is possible to know the prototypes by means of the specific cases. “But as it is we use variables, that is to say we talk, so to speak, of the prototypes by themselves, quite apart from any individual cases” (65) (emphasis in original).

2. Prototype cannot readily be defined but gradually formed out of experiences. With “getting up in the morning” as an example, Wittgenstein remarked, “the prototype of the act of volition is the experience of muscular effort” (Wittgenstein 1958/1969, 151). The problem is that it is impossible to define the experience of muscular effort. Again, with “getting hold of the expression of this face” as an example, Wittgenstein (162) wrote that seeing the expression of the face is not to find the prototype to which it corresponds. It is not that there had been a mould in our mind and the picture we see had fallen into that mould, but that “we let the picture sink into our mind and make a mould there” (163).

3. Prototype should not be confused with “object.” Wittgenstein pointed out that we tend to ascribe what is true of the prototype to the object because we want to give the characteristics of prototype a foothold. But “since we confuse prototype & object we find ourselves dogmatically conferring on the object properties which only the prototype necessarily possesses” (Wittgenstein 1977/1998, 21). The prototype must characterize the whole approach and determines its form, not any single case.

4. Prototype cannot be readily found for subjective experiences. Wittgenstein (1953/1958, 439) asked, “In what sense can one call wishes, expectations, beliefs, etc. ‘unsatisfied’? What is our prototype of nonsatisfaction? Is it a hollow space?” These questions indicate that it is rather difficult to look for prototypes of lots of subjective experiences.

These four ideas demonstrate that prototype in the Wittgensteinian sense does not refer to any individual case or the so-called “best exemplar” at all. This has in fact been verified by empirical studies, among which Solso and McCarthy (1981, 499–503) needs to be briefed here. In Solso and McCarthy’s study, three male faces were drawn at the beginning and assumed to be the prototype, which bore four distinctive facial features: hair, eyes, nose plus chin, and mouth. A face bearing one of the four features has a degree of similarity of 25% with the assumed prototype, for the degree of similarity with the prototype face was determined by the four features. Four levels of exemplars were then derived, each representing a different degree of similarity to the prototype (75%, 50%, 25%, and 0%, respectively). Reproductions of these faces were collected and put in a booklet that contained 10 faces: three 75% exemplars, four 50% exemplars, and three 25% exemplars. In the first of the two experiments, 36 subjects were asked to look at faces in the booklet which does not include the prototype face and try to memorize them. After a five-minute distracter task, the subjects were asked to look at faces in the test booklet, which contained the prototype face, four exemplars shown in the first experiment (one 75% exemplar, two 59% exemplars, and one 25% exemplar), and four exemplars which were not shown to subjects before (two 50% exemplars, one 25% exemplar, and one 0% exemplar). The experiments were repeated six weeks later. The results showed that “35 out of 36 subjects falsely recognized the prototype as an old item, which was, in fact, a new item” in the test booklet (500). This study was aimed to pursue pseudo-memory, but it led to insights into the understanding of prototype. It provides strong evidence for Wittgenstein’s thinking on prototype.

Prototype in the Wittgenstein sense, therefore, has four typical features which are in accordance with high-order cognitive process that is dynamic. The first fundamental characteristic is abstractness. That is to say, prototype should be more than “the original model or form.” It is something abstract. Empirical studies on perception, for instance, have proved that prototypes in perception are abstract and do not refer to specific cases (see Sternberg 2003).

Research (e.g., Solso and McCarthy 1981) also indicates that prototype in a certain space–time can be formed in people’s mind out of a large number of exemplars. Note that by space–time we mean any configuration of contexts through dimensions of space and time. Prototype formed in the mind usually integrates the most characteristic features of a form even if people have never seen any exemplar that owns all the features. So prototype is also characteristic of integrability.

As was shown in Solso and McCarthy (1981), an original model was assumed to be the prototype and then a series of models were created according to the degrees of similarity with the original model. In the first test, created models but the prototype were shown to the participants. Then in the second test, the created models together with the prototype were shown to the participants. Most participants (35 out of 36) felt confident enough to tell that they had seen the prototype in the first test. This suggests that once a good exemplar (which bears most of the common features) is shown, it gains the priority of being recognized and kept in mind for a certain period of time.

Since prototype originates from experiences of members of a category and such experiences may change when new members join in, prototype is subject to changing. Once a new member is added to a category, the prototype of the category will be meant in a more or less extended way. Prototype is thus characteristic of expansibility.

3. Translatability viewed from the Wittgensteinian prototype

According to the universalist view, conceptions with heavy culture load are surely translatable. However, terms like “Dao” have not been sufficiently conveyed in any other languages by far. Can we then judge by such reality and say that the monadist view is right? Perhaps not. As is known to all, hundreds of versions of “Dao” have been provided, and constant efforts and evolutionary developments in translating culture-specific works like Daodejing are quite obvious to readers today. These facts show that both the universalist view and the monadist view are reasonable, but neither can satisfactorily explain the whole phenomenon of “Dao” translation. If observed from the perspective of Wittgenstein’s notion of prototype, translatability may be seen as a category which is both subjectively and objectively determined. Similar to prototype in general, translatability should bear four features: abstractness, integrability, priority, and expansibility.

3.2. Integrability

“Integrability” means that the translation of a text may integrate characteristic elements of all possible interpretations and understandings for a certain context (space–time), as long as the elements reflect or just concern the text. Interpretations and understandings involve both the source text (ST) and the target text (TT). Within translation studies, relationship between ST and TT can be roughly drawn as shown in Table 1.

Table 1.

Relationship between ST and TT.

ST-oriented position	TT-oriented position
TT relying on ST	ST relying on TT
Survival of ST determined by ST itself	Survival of ST determined by TT
Significance determined by quality of ST	Significance determined by quality of TT
ST being of fixed identity	ST changing with each time of translating

Literature suggests that more and more scholars began to take the stance that the relationship between ST and TT is not so simple, far from a clear-cut one. As Kujamäki wrote, “Even a source text that seems to present no translation difficulties in surface structure … is still a powerful constraint in translation and very likely to produce language patterns in translation which are alien to or deviant from general target-language usage” (2004, 188).

The relationship between ST and TT shows quite a different picture if viewed from the perspective of Wittgenstein’s notion of prototype. Interpretations for and understandings of both the ST and the TT serve as exemplars, and all characteristic elements of them can be integrated together to form a typical exemplar in a certain space–time. Here we may turn to “Dao” for better understanding of “integrability.” In the history of understanding Daodejing, the text “came down to us in various editions to which were typically attached commentaries of various kinds” (Penny 2008, xiv). Among the Chinese editions, the most well-known are the Guodian text, the Wang Bi version and the Mawangdui version. “Dao” in these texts and others signifies “way,” “path,” “route,” “doctrine,” “principle,” “primordial essence,” “fundamental nature,” “natural order,” “ineffable qualities,” “truth,” “origin,” etc., in the Chinese context. The English translations add to the understanding of “Dao” such interpretations as “Providence” (E.H. Parker in 1903; see BPS 2014), “Reason” (D.T. Suzuki and Paul Carus in 1913; see BPS 2014), “the doing” (Tom MacInnes in 1927; see BPS 2014), “IT” (Edwin Denby in about 1935; see BPS 2014), “Way-making” (Roger T. Ames and David L. Hall in 2003; see BPS 2014), and “the Cosmic Consciousness” (Brian Donohue in 2005; see BPS 2014). Some versions of Daodejing “are serious attempts at making sense of a famously difficult text, but many more have little to do with the original” (Penny 2008, vii). However, all interpretations and translations are already integrated into people’s understanding of “Dao.”

Generally, the understandings and interpretations of the ST may form up a set of exemplars and those of the TT may form up another set of exemplars, and still those potential understandings and interpretations come up with a set of potential exemplars (see Figure 1). The common core of these sets may be regarded something suitable to all, and that may be the prototypical elements with which future translations may directly integrate.

Figure 1.

Prototypical elements formed out of integrations of three sets of exemplars.

In Figure 1, we can see that the formation of prototypical elements is partially determined by the set of potential interpretations and understandings. The potentiality indicates that a text with high literariness is always open to be translated and retranslated. That is the reason why we have so many translations of Daodejing and many other translations are still on the way to come to literature.

4. Judging the translations of “Dao” from the perspective of Wittgensteinian prototype

With so many translations of a text available, how do we go on to judge or choose what might be a “good” translation? Early in the eighteenth century, Tytler (1791/1907, 8–9) took a good translation to be:

That, in which the merit of the original work is so completely transfused into another language, as to be as distinctly apprehended, and as strongly felt, by a native of the country to which that language belongs, as it is by those who speak the language of the original work.

Good translation judged in this way already takes subjective experience into consideration. Translation is “probably the most complex type of event in the history of the cosmos” (Richards 1953, quoted in Nida 2001, 3), for it concerns not only material factors but also experiential factors. Experiential concepts, as Wittgenstein found, are rather difficult to be grouped under a certain category and the only way to observe them is to consider the family resemblances of the members of a category. That is to say, they would be better treated in terms of Wittgensteinian prototype. As was discussed above, abstractness, integrability, priority, and expansibility are the four features of prototype in the Wittgensteinian sense. When we judge what is the good translation of “Dao,” we may focus on these features. Exemplars such as “a Way (Dao)” by Moran (1993; see BPS 2014), “the Way (Tao)” by Sarbatoare (2002; see BPS 2014), and Tao (Truth) by Org (2000; see BPS 2014) are abstract, integral, and expansible to some degree. Other translations such as “The TAO, or Principle of Nature” by Frederick Henry Balfour (1884; see BPS 2014) are not terse, but they seem to be intended for some degree of integration.

In some contexts, a translation may be a good one if it serves its purpose. For example, the translation of Daodejing by Stanislas Julien in 1842 was known as the best translation by most sinologists (see http://www.guoxue.com/?p=10865). One of the reasons is that Julien’s translation was the first complete translation of Daodejing into a western language. Yet, older translations may become “classic” in their own right (Armstrong 2008, 17). Later in 1911, Richard Wilhelm published his German translation of Daodejing, and it was regarded as the translation which is closest to the original (Song 2012). This is perhaps due to Wilhelm’s experience in China. He worked in China from 1899 to 1921 and became a follower of Confucius thinking. However, according to Brockhaus Enzyklopädie in 1923, the German translation of Daodejing by Victor Von Strauss was considered the best German translation (see Chen 2004). These instances show that the quality of translation depends on the purpose it serves.

The cultural background is crucial to lexical choices in translation and it has big impact on judging the translations. For example, we examined 129 English translations of Daodejing which are clear in translator identity: 23 translators were native Chinese speakers and 106 translators were westerners. Among the 23 translators, only two used words other than “Dao” or “Tao” to translate “Dao” (about 8%). By contrast, among the 106 western translators, 46 used words other than “Dao” or “Tao” (about 43%). Clearly, translators with Chinese background may consider “Dao” or “Tao” the best translation of “Dao,” and non-Chinese translators may not. This implies that the judgment on what is the best translation also depends on the cultural background. This resounds with the idea that judging good translation should be guided by translation NORMS, rules and conventions which are culture-specific (Chesterman 1993; Toury 1995).

Among the English versions of “Dao,” many are “Way” and “Tao.” “Way” may be a good translation, for it can mean “courses of life or God’s providence,” law of nature, or commandments of God. But it is generally agreed that “Way” in the western culture does not indicate “absoluteness” or “nothingness” as “Dao” in Chinese culture does. “Tao” seems to be equivalent to the original, but it cannot be regarded as a good translation before target readers are very familiar with the source language and culture. At present, some translations which are abstract, integral, and expansible may be taken as the good translations, e.g., Tao (Nature), Way (Tao), and Dao (Law). In other words, translations which bear most or all of the four features may be treated as good exemplars. In any case, “Dao” will be fully conveyed some day in the future as long as intercultural communication continues. When “Dao” is known as a lexical item in daily life by most English readers, the translation may not only “stand on its own and compete with”, but already “surpass — the source text” (Baer 2011, 10), as long as English culture is the dominant one. On that day, people may search for the etymological origin of the “English” word “dao” (cf. the word “logic” in English).

5. Conclusion

Translatability is dynamic because new TT can always be added to the formation of the category, which accords with the expansibility of prototype. Either through explicitation which starts from a general meaning in ST to a specific meaning in TT, or through implicitation which starts from a specific meaning in ST to a general meaning in TT (Klaudy and Károly 2005), exemplars add to the formation of a concept, a sentence, or a text, leading to the ever expanding of a classic. Translatability is also an issue of evolution because the subjective elements of ST and TT are evolving all the time. “From the point of view of the target literature, all translation implies a degree of manipulation of the source text for a certain purpose” (Hermans 1985, 9). This kind of manipulation can only be applied to the subjective aspects, not to the material texts. In other words, translatability itself is developing and evolving along with exemplars of ST and TT. If to put translatability into the whole human course of space–time, then everything is translatable.

Translatability is abstract and integral, and it gains priority in recognition and categorization in certain cases. Every translation, whether good or poor, are encouraged to be publicized and then integrated so that translatability can evolve into more and more satisfactory stages. On such stages, translators and readers all contribute to the formation of different exemplars of translation as long as they air their views related to the text. These exemplars can be integrated, and they possess the potential to become the fundamental meanings for translations in certain space–time. Those being recognized first are more likely to become typical exemplars, and there will be changes in the degree of translatability when new members join in. That’s why classics such as Daodejing are always waiting to be retranslated and re-surpassed.

We agree with Pym (2004, 127) when commenting on Hjelmslev’s thinking on translatability in saying that “Emphasis should instead be placed on what Hjelmslev cites from Kierkegaard as the possibility of 'working over the inexpressible until it is expressed'”. The inexpressible may be expressed after all, with exemplars of all possible kinds at hand.