(Ernst-August Gutt, Summer Institute of Linguistics)

 

Abstract

This paper builds on the relevance-theoretic account of communication. It attempts to show that the question whether a textual feature of the original should be represented in the translated text as a "communicative clue" depends on a) the intentions of the original communicator and b) the translator's notion of what his or her task is. Regarding the impact of the translation, it also depends on how well the translator's intentions match the expectations of the audience.

 

The nature of communication

This study builds on the conviction that human communication essentially relies on inference. More specifically, I start from the relevance theory of communication developed by Sperber and Wilson (1986/1995). According to this theory, communication can and often does involve the use of codes, such as human language, but ultimately inference overrides coding. This is part of our everyday experience. We often can correct a communicator's slip of the tongue on the basis of "what makes sense". "Making sense" crucially involves inference. Thus even if a person asks us quite distinctly to pass them the "toffee", under certain circumstances we may be almost completely sure that they actually want coffee and that's what we'll pass them. Thus inference can override what is linguistically encoded.

 

Textual properties and Communicative clues

According to inferential accounts of human communication, a text is a verbal stimulus designed by the communicator in such a way that the audience can infer from it what the communicator intends to communicate. Put a different way, the communicator builds properties into her text that will lead the audience to the intended interpretation. In Gutt 1991 such properties were referred to as "communicative clues" (Gutt 1991:127). As was pointed out there, the notion of "communicative clue" is not essential for a relevance-theoretic account of translation. On the other hand, this notion has been found helpful by a number of people and there was a request to say more about it here. The main purpose of this paper will be to emphasise that for "communicative clues" to be helpful tools, other aspects of the whole communication process need to be carefully considered.

The "meaning" of the original text

One basic point the translator needs to remember is that the notion of the "interpretation intended by the communicator" in itself demands careful consideration. On the one hand, there is not necessarily a sharp dividing line between what a communicator did and did not communicate; rather, information can be communicated with varying degrees of strength, moving along a cline from strongly communicated to not communicated at all, with no breakpoint in between.

On the other hand, the expression "communicated" needs to be handled with care. The primary focus of relevance theory as developed by Sperber and Wilson (1986, 1995) and as applied to translation in Gutt (1991) is a particular kind of communication, called "ostensive-inferential communication", sometimes shortened to "ostensive communication". This is probably the richest form of communication in that, as Sperber and Wilson point out, it "may have social implications that other forms of information transmission do not [have]" (Sperber and Wilson 1986:62). It not only transfers information, but creates a mutual awareness between the communication partners of what has been communicated between them. It brings about "… a change in their possibilities of interaction and in particular, in their possibilities of further communication" (Sperber and Wilson 1986:61f). Information I have ostensively communicated to you, I can assume as shared between us from then onward.

Here is an illustration of the potential social difference between awareness and mutual awareness. This is an incident the author experienced in a restaurant. One of the guests ordered soup with a sausage in it; when trying to take the sausage out of the bowl onto his plate, it slipped and dropped on the table. The author noticed it but quickly looked away. When the man had managed to pick up the sausage from the table and put it on his plate, he looked around if anybody had noticed his mishap. The author gave no indication whatever that he had noticed.

Now, he could have done so, perhaps by a brief smirk on his face, or by staring at the spot on the table where the sausage had dropped. Had he given such an indication, that mishap would have become part of the mutual cognitive environment of the author and the other guest, and most likely that would have changed their relationship: realising that the author had noticed that the guest had dropped the sausage on the table, and also that the author had noticed that he had noticed that the author had noticed this, there could easily have been embarrassment between the two persons. In the event, the author's discretion prevented a mutual awareness and there was no embarrassment. So, mutual awareness can make a difference to social relations.

However, as Sperber and Wilson point out, information can be transferred between people without the necessary creation of mutual awareness. Suppose the communicator happens to feel angry with her audience. Though she might not want to show her anger and hope that it went unnoticed, the audience may still have picked it up, perhaps from voice quality, intonation, facial expression or body language. In such cases the audience would be aware of the communicator's feelings, but the communicator would not have intended them to be; she might not even realise that they did pick up her feelings. Hence there would have been a transfer of information without the establishing of mutual awareness. In fact, the communicator could have denied her anger, if questioned.

The amount of information conveyed incidentally rather than ostensively can be large. One can often tell from the way a person talks what educational background they have, the social group they belong to, what kind of audience they believe they are talking to, perhaps the geographical region they come from, their emotional state, whether they are short-breathed, have a cold or not, and other kinds of information. Very often the communicator may neither intend to convey all this information nor even be aware of its accessibility. How much of this information the audience actually picks up will very much depend on their mental alertness, intellectual capabilities and experiences.

It may seem tempting to set up some kind of typology of information to distinguish what is ostensively communicated from incidentally transferred information. We might, for example, assume that the semantic contents of verbal expressions are usually part of what is ostensively communicated, but that the more "associative" aspects of meaning - register, connotation, dialect etc. - fall under incidental information.

However, such a typological approach would not work. Take, for example, a novel where the writer portrays one of his characters speaking a certain sociolect. At the level of the writer, the features characterising that sociolect are certainly meant ostensively: the reader is meant to notice those features and to use them in his interpretation of the novel. At the level of the story itself, however, that is, from the viewpoint of the person in the story, these features would be incidental. Since the same set of properties can be ostensive at one level, and incidental at another, a typological distinction between ostensive and incidental textual properties will be doomed to failure.

The distinction rests rather upon the intentions of the communicator. It is, therefore, part of the communicator's task to make her intentions in this regard mutually manifest, so that the audience can arrive at the intended interpretation.

As always, the accessibility of the right contextual information plays a key role for inferring the communicator's intended meaning. Imagine the following situation. A colleague comes to you and tells you, in a broad Yorkshire accent, that the manager of the company wants to see you. Now let us assume two different cases.

Case A: It is mutually known by you and your colleague that he is from Yorkshire and that he always speaks with this heavy accent.

In this case, you might notice the accent - though you might not, if that is the way the person normally talks - but assume it to be incidental and of no significance to the intended meaning of your colleague.

Case B: It is mutually known by you and your colleague that he normally speaks standard English, but that he is good at imitating dialects. It is further mutually known that the manager speaks with a broad Yorkshire accent.

In this case, you would probably notice the accent; knowing that your colleague does not usually speak with that accent, you would have to conclude that he is using it intentionally on this occasion, and you would look for its relevance. Since he mentioned the manager, information about that person would be highly accessible, and you could reasonably expect that your colleague expects you to use that information for interpreting his message. You would probably interpret his use of the accent as a parody, either friendly-funny in intent, or possibly mocking or derogatory, depending on your colleague's relation to the manager. Of course, the colleague may give further evidence of his intentions by some facial expressions, but not necessarily so.

Thus, what is acoustically or phonetically the same utterance, with the same textual properties, can lead to rather different interpretations, depending on the mutually manifest context, which is important for working out which of those properties are to be taken as "communicative clues".

Note that in case B it is possible that the utterance of your colleague is practically identical with what the manager actually said to him. So your colleague could not be accused of having distorted the manager's message by inaccurate reproduction. There is a distortion, but it resulted from the accurate reproduction of incidental features, which were irrelevant to the intended meaning of the original message, and therefore must be relevant in some other way.

If one were looking only at intra-lingual communication, one could simply say that communicative clues are a subset of the textual properties that are significant for the intended meaning. There would not be any difference in essence between a textual property and a communicative clue.

 

Communicative clues in translation

However, the situation changes when considering cross-lingual communication, and this is where it seemed helpful to form a more abstract concept than textual property. The reason is that languages differ in the inventory of linguistic features or properties they have; hence property A of language X may simply not be found in language Y. Nevertheless, one can very often find some means B in language Y that achieves the same or at least similar effects as property A did in language X, assuming identical contexts. Properties that can be linked in this way are referred to as corresponding "communicative clues".

Take, for example, the pronominal distinction of gender, which is so common in Indo-European languages but does not exist in the Finnish language. This lack of gender distinction in Finnish can sometimes be compensated by the use of a noun that does include information about the gender of the referent. Thus, at a place where the gender marking in the English pronoun she is important, one could, for example, use the Finnish noun äiti '(the) mother' which would make the gender clear, instead of the gender-neutral pronoun hän.

 

The handling of "communicative clues"

Turning to translation itself, here are a couple of examples.

 

A premier's resignation

The following example is given by Brain Mossop (1987) in his article "Who is addressing us when we read a translation?" He compares two translations of a note, written in French by René Lévesque, former Prime Minister of Quebec, on the occasion of his resignation.

Version A: I would appreciate if you could transmit for me to the National Council this simple message: Thank you from the bottom of my heart, thanks to you and to all those, who will recognize themselves, and who have not stopped for so many years paying with their selves and their pocketbooks in order to build, implant, maintain this project which is so healthy and democratic and which we have designed together for our people.(Mossop 1987:11)

This English translation was given in the Globe and Mail newspaper. Version B was apparently produced by Mossop himself to illustrate the "idiomatic" type of translation:

Version B: I would appreciate if you could transmit this message to the National Council  for me: Thank you from the bottom of my heart. My thanks to you personally and all those men and women, - they know who they are -  who have for years been devoting their energies and their pocketbooks to the task of building,  and maintaining a healthy and democratic road to the future, a road which together we have laid out for the people of Quebec. (Mossop 1987:11f)

Interestingly, when evaluating the two versions, Mossop prefers version A because it avoids two problems he sees with version B. In the first place, "… Canadians will have heard Lévesque speaking English on television and they will know that … Lévesque does not speak English idiomatically" (Mossop 1987:12). Therefore, such "readers might wonder whether the 'voice' they are 'hearing' [in version B, that is E-AG] is that of the same person they have heard on television" (Mossop 1987:12). Version A, by contrast, "… captures (perhaps not always successfully) Lévesque's thought process - not just the force of what he says - and by its unusual language it avoids assimilating Lévesque to English-Canadian culture" (Mossop 1987:12).

The second problem is, according to Mossop, that "… this resignation is a historical document, and its translation should therefore indicate something of the author's personal style" (Mossop 1987:12).

 

Partial analysis using "communicative clues"

Applying the notion of "communicative clues" to this example, we could speak of the differences between versions A and B as differences in communicative clues. More specifically, these clues are present in version A, but missing in version B. The table below provides a summary of these "clues" and their significance, as perceived by Mossop.

 

Clues in version A (lost in version B)	Significance/relevance
1 who will recognize themselves	Odd expression in English
2 implant	Emotional involvement
3 'build' and 'maintain' have an object that does not 'go with' them	Odd expression in English (unusual collocation)
4 paying with their selves	Emotional involvement (notion of self-sacrifice)
5 juxtaposition of three clauses without conjunction	Abnormal grammar
6 our people	Emotional involvement (identification)

Clues (1), (3), and (5) are all "odd" English; clues (4), (2), and (6) might be said to make version A emotionally more expressive than version B, though (4) may also be judged at least unusual for ordinary English.

 

Comments on the "message" of version A

As mentioned earlier, "communicative clues", are properties built into the text to guide the audience to the intended interpretation. Clues (1), (3), and (5) lead the audience to infer that the English of the communicator was odd. According to Mossop, this is the very inference that should be conveyed, in order to avoid "assimilating Lévesque to English-Canadian culture" (Mossop 1987:12).

The problem that Mossop does not discuss, however, is that the original message was in French, not English, and that it was perfectly idiomatic French. Hence if there was anything that the original audience could infer about the speaker's language competence, it was that he mastered French; there was no clue about his abilities in English whatsoever. It is therefore clear that the communicative clues (1), (3) and (5), giving evidence about Lévesque's problems with English, were introduced by the translator alone; they had no base in the French original, and certainly did not support its originally intended meaning.

 

Evaluative comments

How can one evaluate such a practice? The most obvious question to ask is whether such practice is legitimate for a translator qua translator to carry out. This question is notoriously difficult to answer because it all depends on what one takes translation to be, and there simply is no consensus on this concept. What one could do is try to find out which notions of translation would support such practice, and which would not. Though an interesting study of an encyclopaedic nature, it would not necessarily help any translator deal with the issue.

Another, perhaps more useful, evaluation - and this is the one to be pursued here - could examine how successful the translator would be likely to be in communicating this particular information to the audience. Since that is the point of providing such clues, this kind of evaluation seems to be only fair. So the question to ask is what kind of inferences the audience would draw from these clues. Like with all inferences, the premises, that is, the contextual assumptions, that enter into them are of considerable importance. So let us examine the influence different contexts would be likely to have on the interpretation of version A.

 

Case 1

Assumed context: The knowledge of Lévesque's problem with speaking idiomatic English is highly accessible in the audience's mind.

Interpretation: When reading version A in the newspaper and noticing the unidiomatic expressions, they would most likely infer that this was another instance of Lévesque speaking English. That is, they would probably consider the text to be a verbatim quote of what the politician had said or written in English.

Evaluation: The translator's intention of reminding the audience that Lévesque speaks unidiomatic English would have been fulfilled, but it would have been accompanied by a wrong conclusion about the circumstances of this act of communication.

 

Case 2

Assumed context: Like in case 1, the knowledge of Lévesque's problem with speaking idiomatic English is highly accessible in the audience's mind. In addition, they happen to know that Lévesque wrote his note in French, and that therefore the newspaper version must be a translation.

Interpretation: When reading version A against this context and noticing the unidiomatic expressions, the audience could only attribute these expressions to a translator; attribution to Lévesque would be excluded by their knowledge that he wrote this message in French.

Turning therefore to the translator's involvement, two different lines of further interpretation would be possible. One line would start from the assumption that the unidiomatic expressions were unintentional, that they were slip-ups by the translator. This would probably lead to dissatisfaction with a seemingly incompetent translation. In this case the translator's intention of using the clues to remind the audience of the politician's poor English would be lost.

The other assumption could, however, be that those expressions were intentional. That is, the audience could assume that the translator could have written idiomatic English but purposely chose to deviate from it. Processing non-standard language expressions requires more effort than processing standard ones, and so the audience would expect the translator to have intended to communicate to them more than standard language could have done. (This follows from the principle of relevance, believed to be operative in ostensive-inferential communication.)

What could this extra information be? Again, there would be different options. One option would be to assume that the non-standard language was necessary to bring out subtleties in the meaning intended by the original speaker which would have been lost by idiomatic expressions of English. People familiar with translation matters would probably make this assumption quite readily. In the current text, however, no such additional bonus-meaning seems to be available: virtually the same contents could have been communicated in natural English.

So, if the reward for the additional processing effort did not lie in a richer understanding of the original meaning, then what other additional meaning could the unidiomatic expressions serve to provide? At this point the audience would perhaps take a wider look at the whole communication process, thinking of the speaker, the surrounding circumstances etc. They might notice that the unidiomatic expressions resembled those used by Lévesque when speaking English. Hence they could conclude that the translator had used the non-standard expressions to draw attention to Lévesque's inadequate mastery of English. (Note that ordinarily these interpretation processes operate sub-consciously; we are not usually aware of them.)

Whether they would accept this interpretation as the intended message from the translator would partly depend on their notion of "translation". As a general policy, it seems rather odd to expect that the translation of someone's speech should reflect the original communicator's knowledge of the target language. This could lead to ridiculous or even unintelligible renderings, depending on the original speaker's ability in the target language, and one wonders what the translator should do if the original speaker had no knowledge of the target language at all. In many contexts it would also raise serious questions of propriety since the deliberate imitation of mistakes in someone's speech tends to be shunned by society as mockery.

At the same time, it seems the language ability of politicians in Canada was a topic of public interest. Perhaps this is the key to Mossop's concern. Version B appeared to present Lévesque speaking idiomatic English - which was contrary to fact and therefore, Mossop felt, had to be corrected. Presumably there were others who shared his concern. The size of this group would determine the degree of success such a translation could count on in the receptor community.

 

Comments on the "message" of version B

Turning to version B, we are dealing with communicative clues (2), (4), and (6), all of which were arguably present in the original. In number (2), the metaphor of enraciner, 'implant', is lost altogether. The rendering devoting their energies is emotionally less expressive than the original payer de leur personne, and pour notre peuple clearly shows that the speaker identifies himself with the cause of Quebec, whereas for the people of Quebec fails to show this identification.

What is interesting in this particular example, is that version B is Mossop's own work, and one wonders why in version B he combined the choice of idiomatic English with the omission of those expressive clues. This is certainly not a necessary combination - one could well imagine a version C which did use idiomatic English while at the same time maintaining the clues to the emotional involvement of the original speaker.

From what Mossop says, his choice seems to have been strongly influenced by two distinct stereotypes: one was that of a French-speaking politician with certain strong sentiments for Quebec and a French accent, and the other of an English-speaking politician, leading one of the English-speaking provinces, who would have idiomatic English but no strong sentiments for Quebec.

I think within the scope of this paper, we shall have to leave it at that, though it would no doubt be fascinating to further explore the intricacies of political sentiments in Canada, which we find exerting subtle influences even on everyday translation matters.

 

A look at the Schocken Bible

In 1995 a new translation of the Bible into English, the Schocken Bible, appeared, prepared by Everett Fox, a professor of Jewish studies (Fox 1995). In the words of Natalie Weinstein, Fox believes that "… too many translations already try to "spoon-feed" the Bible by using everyday, modern English that makes the reader feel comfortable. In Jewish tradition, however, studying Torah isn't supposed to be easy." (Weinstein 1997) So, Fox himself characterises his purpose as follows: "Our approach is that the Bible is difficult and something that must be wrestled with. (…) This tries to give … a taste of what's there." (Weinstein 1997)

 

The approach

Fox's approach was inspired by the translation of the Bible by Martin Buber and Franz Rosenzweig into German. These Jewish scholars were disturbed by the fact that the existing translations of the Bible were "pulling the 'contents' of the text over into another language, not necessarily abandoning a priori the peculiarities of the elements, structure and dynamics, but abandoning them too easily where the brittle 'form' seems to hinder the passing on of the contents" (Buber 1954:4; translation my own, E-AG). To this practice Buber objects: "As if a genuine message, a genuine saying, a genuine song contained a 'what' that could be detached from the 'how' without damage, …" (Buber 1954:4; translation my own, E-AG). No message can be completely transferred from one language to another, but the translator must try to approximate as best he can, "getting as close as the limits of the language into which he translates will allow; but to these limits the translator needs to penetrate" (Buber 1954:7; translation my own, E-AG).

In this spirit, Fox translated Genesis 1,1-3 as follows:

At the beginning of God's creating of the heavens and the earth when the earth was wild and waste, darkness over the face of Ocean, rushing-spirit of God hovering over the face of the waters -- God said: Let there be light! And there was light. (cited in Weinstein 1997)

One of the major principles that would distinguish Fox's translation from most other contemporary translation efforts is his explicit commitment to concordant translation, because, following Buber and Rosenzweig, he believed that the repetition of the same Hebrew word was important for making connections or for establishing a theme.

For example, forms of the word avod, or "serve," appear throughout the book of Exodus. The Israelites must serve the Egyptians; they become "serfs," which is the word Fox uses instead of the traditional translation of "slaves." Later in the story, Moses uses the root word "avod" in another context, asking the Pharaoh to free the Israelites so they can serve God. That wording, Fox said, shows the transition of the Israelites from serving humans to serving the Divine. (Weinstein 1997)

So here is another instance where the translator uses unidiomatic expressions in the target language as "communicative clues". Yet there are some important differences to Mossop's case. These differences arise from differences in contextual knowledge.

First of all, it seems highly unlikely that the target language audience would assume that the original might have been written in English by someone with a less-than-perfect command of the language. More likely, English readers would naturally attribute the noticeable "oddities" of English to the translator, and they would again have the choice of taking them either as accidental slips or as deliberate. Since Fox is rather explicit in explaining his purposes, the careful reader would have to accept them as deliberate, and it would also be clear that the oddities are meant to contribute to the intended interpretation rather than provide information about the original writer.

 

Evaluative comments

What about the prospects for the success of Fox's translation - success in the sense that it will convey to his audience what he intended to communicate to them through it?

From a relevance-theoretic point of view, the use of unidiomatic expressions means that the translated text demands more processing effort from its audience than a rendering using the words, grammar and idioms of English in their normal way. In order for Fox to succeed, the audience must a) be able to make sense of the translated text, b) have access to the contextual information which Fox had in mind and which will yield the promised gain in understanding, and c) be willing to invest the additional effort needed for a) and b).

Interestingly, the expenditure of this additional effort is one of the main effects Fox aims at; as he said, he wants to give the English reader "a taste" of the fact that "the Bible … must be wrestled with" (as cited in Weinstein 1997).

Reactions seem to indicate that some people meet these conditions and share Fox's taste. Thus David Noel Freedman, a professor of Hebrew biblical studies and editor of the Anchor Bible Series, comments: "It's a very fresh approach. I'm surprised it worked" (as cited in Weinstein 1997) Similarly Nahum M. Sarna, a professor emeritus of biblical studies, who also worked on the Jewish Publication Society's revised translation, writes: "It's extremely useful conveying, to those who don't know Hebrew, the sense of the original" (as cited in Weinstein 1997).

Others apparently fail to meet those conditions. Rabbi Chaim Stern, himself a translator from Hebrew into English, finds that it is "not a workable proposition. … English doesn't like that kind of structure. It isn't native to English. It's native to Hebrew." (as cited in Weinstein 1997) It seems to me that Stern has a point here, if one considers the following. Granted that the interpretation of the Torah is not an easy matter, even in the Hebrew original, there is however, an important difference to Fox's English version because in his translation the difficulty is often linguistic, resulting from unusual or even ungrammatical English constructions; for the Hebrew original, however, that is certainly not the main problem, though linguistic problems do arise from textual corruptions and the like.

Furthermore, one suspects that, contrary to Sarna's evaluation, it will mostly be readers who already know Hebrew who will recognise Hebrew structure behind the distortions of English, but surely they are not really the ones who will need Fox's translation for that experience.

Robert Alter, professor of Hebrew and comparative literature, calls Fox's text a "bold and admirable" effort but objects that "The English too often seems strange", an impression which he sees in conflict with the "magnificent, literary Hebrew" of the original (as cited n Weinstein 1997). This comment shows, that for this critic the approach is mistaken because it fails to lead to the conclusion that the original version of the Bible was of high literary quality.

That is also a concern when considering the purpose of preserving communicative clues present in the original: the Hebrew writers certainly did not compose their texts in order to communicate an awkward use of language. So, again it seems that the clues the translator chose are not in line with the intentions of the original author.

So, one suspects that the group with whom Fox succeeds will be small compared to those who prefer English translations which are more readable, even if they lose some aspects of the original meaning. It is difficult to estimate what the overall balance of gain and loss is, and Fox himself concedes that "… his method sometimes loses the nuances found in other translations" (as cited in Weinstein 1997).

 

Lessons from the examples

What can we learn from this study with regard to "communicative clues" and their use in translation?

First of all, in translation "communicative clues" need to be considered at two different levels: the level of the original communication process and the level of the translator communicating with her target audience.

Secondly, the identification of "communicative clues" at neither level is necessarily clear-cut or simple. This is partly due to the nature of ostensive communication, which allows for indeterminacy, and partly due to the practical problem of working out the particular intentions of the communicator in a given instance of communication, especially when there are differences in the assumed contextual information.

Thirdly, the relationship between these two sets of communicative clues touches on the essence of translation. Most contemporary translators would probably agree that their primary, if not their only task, is the expression of clues found in the original that will help the target audience recover the originally intended meaning. Other translators would want to include communicative clues that give access to information about the communication act - e.g. about the speaker or the nature of the original language - even if these clues were not intended or not even present in the original.

Fourthly, the preservation of communicative clues from the original may increase the processing effort, for example, by leading to unnatural or unidiomatic expressions in the target language. The translator will have to carefully gauge whether for her particular target audience this additional effort will be rewarded by additional benefits. Where this is not the case, the translator has no reason to expect success in her communicative effort with that particular audience.

In conclusion, this means that the notion of "communicative clue" cannot be used in translation in any mechanical way. It requires a good understanding of the inferential nature of communication. I therefore suggest that the translator should check her purpose in translation against the following three questions:

1)     What does the audience expect from the translation?

2)     Do they have the contextual information (background knowledge) required for the intended interpretation?

3)     Can they be expected to invest the effort necessary for processing the translated text?

 

References

Buber, Martin 1954 'Zu einer Verdeutschung der Schrift'. Supplement to Buber and Martin 1954.

Buber, Martin and Franz Rosenzweig 1954 Die fünf Bücher der Weisung, Köln und Olten: Jakob Hegner.

Fox, Everett 1995 The Schocken Bible: Volume I -- The Five Books of Moses;, San Francisco Jewish Community Publications Inc., dba Jewish Bulletin of Northern California.

Gutt, Ernst-August 1991 Translation and relevance: communication and cognition Oxford: Blackwell.

Mossop, Brian 1987 'Who is addressing us when we read a translation?' TextconText  vol. 2, pp. 1-22.

Sperber, Dan and Deirdre Wilson 1986 Relevance: communication and cognition. Oxford: Blackwell (2nd edition 1995).

Weinstein, Natalie, 1997 'New Torah translation: a radical approach' (e-mail communication, July 1997)