Translating Presuppositions

1. Introduction

This article discusses some properties of linguistic presuppositions, and some of the issues that arise in translating presuppositions, with a particular focus on Bible translation. One major challenge in addressing this topic is the fact that the term presupposition is used in a broad variety of ways in common, everyday speech, and that many previous discussions of presuppositions in translation have also used the term in a very broad way. ¹

A linguistic presupposition, in the narrow technical sense of the word, is a piece of information that is linguistically indicated to be part of the common ground at the time of the utterance. ² The term common ground refers to everything that the speaker and hearer both know or believe, and are aware that they have in common. By saying that this information is “linguistically indicated” I mean that the speaker’s use of a specific word or grammatical construction signals the shared status of the information. This word or grammatical construction is referred to as the presupposition trigger.

For example, when Jesus says to the Samaritan woman, “Go, call your husband, and come here” (John 4.16), his use of the phrase your husband seems to indicate that she is married at the time of their conversation and that both Jesus and the woman are aware of this fact. Similarly, when Paul writes to the Corinthians, “I made up my mind that I would not make another painful visit to you” (2 Cor 2.1), he implies that he has made a previous painful visit to Corinth, and that the Corinthian Christians remember it. This presupposition is triggered by the word another.

Now in the first example, the presupposition turns out to be false, as the woman herself points out. (Jesus’ reply shows clearly that he knew this. Perhaps he was speaking politely or euphemistically to avoid shaming the woman, or perhaps he was using this strategy as a way of initiating a discussion about her personal life.) This is an instance of presupposition failure.

Another type of presupposition failure occurs when the presupposed content is true but not actually part of the common ground at the time of the utterance. Typically this involves information that is new to the addressee. Such cases are often referred to as informative presuppositions (see section 3 below). A classic example is Jesus’ statement to his disciples at the last supper, “Behold the hand of him who betrays me is with mine on the table” (Luke 22.21). The restrictive relative clause him who betrays me triggers a presupposition that one of the disciples will betray him; but this information was not previously known to them.

In responding to an unshared presupposition like this one, the addressee has two choices. If the presupposed content is uncontroversial, the addressee is likely to simply accept it as new information that can be added to the common ground. This process is known as accommodation. If the presupposed content is controversial or difficult to accept, the addressee may object by saying something like, “Hey, wait a minute, I didn’t know that!” The disciples’ reaction in the example just cited, as recorded in the parallel passages in Matthew and Mark, is of the “Hey, wait a minute” variety. The appropriateness of this “Hey, wait a minute” response is in fact one of the key diagnostic tests for identifying linguistic presuppositions (Fintel 2004).

Presupposition is closely related to information packaging. Because presupposed content is treated as being part of the common ground, Potts (2015) describes it as being “backgrounded,” that is, not part of the “at-issue” content of the utterance (the main point which is asserted in a statement or queried in a question). For this reason it seems intuitively desirable for presuppositions in the source text to be expressed as presuppositions in the target language; this is explicitly argued for by some authors. However, this is not always possible due to linguistic differences across languages.

The present article is organized as follows: section 2 discusses some types of expressions that commonly trigger presuppositions, and some of the standard diagnostic tests used to identify presuppositions. Section 3 discusses types of presuppositions in the source text that can cause difficulty for readers of a translated version. A common strategy for dealing with such cases is to restate the problematic presupposition as an assertion. While this strategy preserves the propositional content of the original, it changes the information packaging by changing content that is taken for granted in the original into “at-issue” content in the translation. Section 4 considers an alternative strategy that involves restating the problematic presupposition as a conventional implicature: content that is taken for granted by the speaker/author, but not implied to be already known by the addressee.

2. Presupposition triggers and diagnostics

3. Presupposition failure and informative presuppositions

In translating a narrative text, the translator must attend to presuppositions on two levels: first, presuppositions triggered within quoted conversations, reflecting the common ground between the participants in the reported speech event; second, presuppositions triggered by the words of the narrator, reflecting the common ground between author and reader.

Explicit rejection of a false presupposition, as in John 4.16, will only occur within quoted conversations. Another example is found in John 9.2, where the disciples ask Jesus, Rabbi, who sinned, this man or his parents, that he was born blind? This is a complex example: the result clause, (so) that he was born blind, presupposes that the man’s blindness was the result of sin; the alternative question, Who sinned, this man or his parents?, presupposes that the sinner was either the man himself or his parents. Jesus’ reply in the following verse rejects both of these false presuppositions: Neither this man nor his parents sinned, but this happened so that the works of God would be displayed in him.

Another striking example occurs in John 7.19-20:

Whereas John 4.16 and John 9.2-3 illustrated the rejection of a false presupposition, in John 7.19-20 we see a false rejection of a true presupposition. The presupposed content (“you are trying to kill me”) is actually true, but the addressees respond as if it were false. John seems to highlight the absurdity of their denial just a few verses later by showing that the plot against Jesus was already common knowledge: Some of the people of Jerusalem therefore said, “Is not this the man whom they seek to kill?” (John 7.25-26).

The other type of presupposition failure, which involves presupposed content that is not actually part of the common ground, may occur either within quoted conversations (as in Luke 22.21) or in the words of the narrator. An example of the latter type occurs in Exod 18.2-3a:

In this passage the temporal clause after he had sent her away presupposes that Moses had sent his wife and sons away; but this information has not been previously mentioned in the text. We might speculate that the original author had in mind an intended audience, of his own time and nation, to whom this fact would have been familiar. But even if this could be shown to be true, it would be irrelevant to the needs of the modern reader, who may find it difficult to accommodate this unshared presupposition. I personally experience a fairly strong “Hey, wait a minute” response every time I read this passage.

Whether or not this particular fact was already familiar to the first readers of Exodus, there are clearly cases where a speaker or author does not assume that presupposed content is familiar to the intended audience. A speaker/author can choose to encode new information as a presupposition, treating it as part of the common ground and expecting the addressee to make the necessary accommodation. A simple example was cited by Stalnaker (1974, 202):

I am asked by someone who I have just met, “Are you going to lunch?” I reply, “No, I’ve got to pick up my sister.” Here I seem to presuppose that I have a sister even though I do not assume that the addressee knows this.

Prince (1978) discussed a pattern that is fairly common in written English, which she called the “informative-presupposition cleft.” In this pattern, illustrated in (6), the speaker/writer uses the presupposition of a cleft sentence to encode information that is known in general (Henry Ford gave us the weekend), but is probably new to the hearer. The hearer is expected to acquiesce in treating this information as part of the common ground.

The term informative presupposition is used to describe information that the speaker/author encodes as a presupposition, as if it were part of the common ground, even though it is actually new to the addressee. While this is a common strategy for expressing uncontroversial content in everyday conversation, it is a potential source of confusion in translated material. We should be very cautious about assuming that informative presuppositions in the SL text will be readily accommodated by RL readers. Each instance needs to be carefully checked.

Consider again the example cited above from John 7.19-20. How is the reader supposed to know that the presupposed content of Jesus’ question (you are trying to kill me) is true, i.e., that the opponents’ denials are false? A sympathetic reader of John’s Gospel may be inclined to give Jesus the benefit of the doubt, but the only explicit indication in the preceding text comes from a reference in John 5.17-18 that involves an informative presupposition:

The words all the more trigger the presupposition that the Jewish leadership had already been trying to kill Jesus. Nothing in the preceding text indicates this, although v. 16 speaks of persecution. Some later Greek manuscripts add a clause to v. 16 that explicitly asserts the content presupposed in v. 18, suggesting that the informative presupposition in v. 18 may have been awkward or hard to accommodate even for readers of the original Greek. Compare KJV, which includes this addition, to NIV, which (like all modern versions) leaves it out:

Another example of informative presupposition is found in the following passage, where the death of Eli’s daughter-in-law is first mentioned within a time adverbial (ESV, NASB, KJV: about the time of her death; NIV: As she was dying) and is thus presupposed rather than asserted. Like the example from Exod 18 discussed above, I find this English rendering difficult to accommodate, or at least highly unnatural, because the information is so significant and the reader is not adequately prepared for it. In contrast, the informative presupposition triggered by the definite NP the women attending her (namely, that there were women attending her) is easily accommodated because it is entirely predictable.

If an informative presupposition in the SL text is not easily accommodated by RL readers, one option is to express the presupposed content as a separate assertion. Some modern English versions adopt this strategy in Exod 18.2:

Several of these versions make a similar adjustment in Luke 22.21:

NLT adopts the same strategy in 1 Sam 4.19-20. NCV here preserves the original time adverbial, but strengthens the meaning of the preceding clause (her pains came upon her), which other versions take to mean simply that she went into labor. The result is to add an extra assertion that is implied but not stated in the original (the mother had much trouble in giving birth) as a way of preparing the reader for the bad news. CEV adopts both measures, strengthening the preceding clause to The birth was very hard and changing the presupposed content to a separate assertion (she was dying).

The direct assertion strategy resolves the discomfort or confusion that readers may feel if they find it difficult to accommodate unshared presuppositions, but the strategy is not cost-free. Stylistically, this kind of restructuring may result in a less vivid or interesting text. Perhaps more importantly, while this strategy preserves the information content of the original, it does not preserve the information packaging of the original. Content that was “backgrounded” in the source text (that is, taken for granted rather than being asserted) is rendered in the RL version as “at-issue” content, the main point of an assertion. Is this an acceptably accurate translation? We will return to this question in the following section.

Of course, much depends on the intended target audience. It is worth noting that NCV, CEV, and NIrV are all intended for readers whose comprehension of written English is on a primary school level. Moreover, there may be considerable cross-linguistic variation in terms of how readily a particular language group can accommodate various types of unshared presuppositions, although I do not know of any research that directly addresses this question.

Another interesting way in which languages may differ with respect to presupposition failure is reported by Matthewson (2006). She shows that speakers of St’át’imcets (also known as Lillooet Salish) systematically lack the “Hey, wait a minute” response to informative presuppositions. Moreover, they do not normally challenge false presuppositions either, unless the presupposed content is totally absurd (e.g., implying that our earth has two suns). She presents a number of examples with the following explanation: ⁶

The following sentences were all offered in “out of the blue” contexts to St’át’imcets speakers. In all cases, the presuppositions failed and were not easily accommodatable. The B utterances in each case are the consultants’ spontaneous responses to A.

[13] Context: Interlocutors all know that Henry is not a millionaire.

 A: “Henry won the lottery again.”

 B: “Oh, good.”

[14] Context: B has just walked into A’s house and there has been no prior conversation apart from greetings.

 A: “Would you like some more tea?”

 B: “Yes.”

Matthewson argues that the difference between St’át’imcets and English in this regard is not (primarily) cultural but linguistic. She shows that St’át’imcets speakers do routinely challenge other kinds of problems, such as contradictory utterances or discourse-initial utterances with unclear pronoun reference. The contrast with English is specific to presuppositions. She concludes:

The data . . . show that St’át’imcets speakers are willing and able to challenge infelicitous utterances of various kinds. I conclude from this that their failure to offer wait-a-minute challenges to failed presuppositions does not result from a cultural prohibition against challenges in general. It must be something linguistic.

Matthewson suggests that, while presuppositions in both St’át’imcets and English are taken for granted (“backgrounded”) by the speaker, they are assumed to be part of the common ground in English but not in St’át’imcets, since the “Hey, wait a minute” challenge is specifically a challenge to the common ground status of the presupposed content. Matthewson argues that the St’át’imcets facts are compatible with the model of presupposition developed by Gauker (1998).

Alternatively, we might suggest that content which is presupposed in English is expressed in the corresponding St’át’imcets utterances as conventional implicature, in the sense of Potts (2005; 2007). Potts defines conventional implicature (a term originally coined by Grice 1975) as content that is part of the conventional meaning of the expression (not a pragmatic inference), is not part of the “at-issue” content of the utterance but is taken for granted (“backgrounded”) by the speaker, and so does not affect the truth conditions of the sentence. We will have more to say about conventional implicature in the following section.

4. Must presuppositions be preserved? Can they be?

Ayman El-Gamal (2001) argues at length that no translation can be considered accurate which does not present SL presuppositions as RL presuppositions and SL assertions as RL assertions. A few selected quotes provide a sense of his views on this issue:

To translate presuppositions as assertions, or vice versa, can distort the thematic meaning of the SL text and produce a text with a different information structure. . . . The importance of preserving SL presuppositions in the [RL] text can hardly be exaggerated. . . . Failure to observe the distribution of information in terms of presupposed, given, or new information would result in some loss of meaning and inconsistency with the intentions of the SL text author.

However, linguistic differences between languages can make it impossible to match certain kinds of SL presuppositions with RL presuppositions. One famous example, from Keenan (1973), concerns relative clause formation. Because Modern Hebrew makes use of the pronoun retention strategy, it is possible to form relative clauses in Modern Hebrew, like (15c), whose literal translation into standard English (15b) would be ungrammatical (indicated by the symbol *): ⁷

The closest equivalent to (15c) in grammatical standard English would be something like (15d). However, as Keenan points out, the speaker’s knowledge of the man is asserted in (15d), but presupposed in (15c). The Hebrew sentence and its English translation convey the same propositional content, but the pattern of assertions and presuppositions is not the same. Thus (15d) fails to be an exact translation of (15c).

McCready (2014) mentions another familiar example, namely the presuppositions triggered by the definite article in languages like English. In at least some contexts, the presupposition associated with this article (existence of a uniquely identifiable referent) cannot be precisely expressed as a presupposition in languages like Japanese and Chinese, which lack a definite article.

We have seen two types of challenges which may cause a translator to render content that is presupposed in the SL as an assertion in the RL: (a) confusion or discomfort on the part of RL readers due to informative presuppositions that are not easily accommodated; and (b) linguistic differences that make it impossible to express certain SL presuppositions as RL presuppositions. At the same time, we have noted that changing a presupposition into an assertion distorts the information packaging of the original text, changing “backgrounded” material into “at-issue” content. Is there another solution?

In some cases, certainly not all, it may be possible to render the problematic SL presupposition as a conventional implicature, in Potts’s sense of the term. Where it is possible, this approach is worth considering, since it preserves the “backgrounded” status of the information without triggering an inference that this information is already known to the addressee.

Some of the core examples of conventional implicature that Potts discusses include non-restrictive relative clauses, appositional phrases, and other parenthetical material. An example of how these constructions might be used to render an informative presupposition is found in the UBS A Handbook on Exodus (Osborn and Hatton 1999, 427–28). One of the alternative translation models suggested for Exod 18.2 makes use of a non-restrictive relative clause, as seen in (16a). This rendering indicates that the author takes the information for granted, preserving the “backgrounded” nature of the information, but does not assume that it is familiar to the reader. The same strategy is adopted in the Indonesian Terjemahan Baru (“New Translation”; 1974), which is generally a relatively formal translation.

As a second example I offer the following possible rendering of 1 Sam 4.19-20, which moves the presupposed content into a parenthetical comment:

To my ear these renderings seem less jarring than the more literal wording of ESV, NASB, and so on, which preserve the informative presuppositions of the original. The suggested parenthetical material indicates that the speaker takes this content for granted (i.e., it is background information rather than the main point of the assertion), but they do not imply that this information is already accessible to the reader.

5. Conclusion

Grammatical and lexical differences among languages can make it impossible to express certain SL presuppositions as RL presuppositions. Moreover, even when it is linguistically possible to achieve this goal, preserving the SL presuppositions may cause problems of comprehension or naturalness for RL readers. This is particularly the case with informative presuppositions, that is, when the content that is encoded as a presupposition is not actually part of the common ground between author and reader at that point of the text.

When translators choose to make adjustments in such cases, the most common strategy is to express the presupposed content as a separate assertion. While this strategy preserves the propositional content of the original in a form that is easier for RL readers to process, it also distorts the information packaging of the original text, changing “backgrounded” material into “at-issue” content. Another strategy that may be useful for some informative presuppositions is to render the problematic SL presupposition as a conventional implicature, for example by expressing it as a non-restrictive relative clause or some other type of parenthetical. This approach, where it can be applied, preserves the “backgrounded” status of the presupposed information without triggering an inference that this information is already known to the addressee.

I am not saying that it is never appropriate to render content that is presupposed in the SL text as a separate assertion in the RL version. Translation always involves trade-offs, calculations of cost vs. benefit for various options in order to balance a variety of competing goals. I believe that there are contexts in which restating SL presuppositions as RL assertions will turn out to be the best option, especially if the goal is to produce a translation that is reasonably comprehensible to the “average” RL reader. My purpose here is to raise awareness of the actual cost associated with that strategy, and to suggest another possible strategy that might be useful in some such cases.