Travaux publics et Services gouvernementaux Canada
Symbole du gouvernement du Canada

Liens institutionnels

Rechercher dans TERMIUM Plus®

Understanding Poorly Written Source Texts

Brian Mossop
(Terminology Update, Volume 28, Number 2, 1995, page 4)

The … translator …, sick of rewriting [source language] SL texts from scratch, building good arguments out of garbage, doing for the writer what the writer should have done for himself or herself, and then getting no credit for it, not even getting mentioned in the publication … may decide to [render] the text in all its ghastliness.

He or she will disambiguate no ambiguous phrasings, silently correct no noun-verb incongruencies, register shifts, or factual errors, prune no repetitiveness, mend no style-context conflicts, rearrange no flaccid sequencing. He or she will be faithful to the letter and the spirit of the text—not out of a fanatic adherence to a principle, of course, but out of an extremely gratifying form of malice, indeed a highly creative and artistically demanding form of malice, in which vengeance is exacted against bad writers through an artful search for just the right degree and shade of [target language] TL verbal shoddiness. [Douglas Robinson, The Translator’s Turn, p. 173]

Part One: Introduction 1

If you are like me, you frequently have a sense, as you begin reading a text submitted for translation, that you are seeing the message only through a glass, darkly. The poorly written source text, like the customer who didn’t send documentation, is one of those ongoing certainties of the translator’s existence. It’s one of those things we’ve been complaining about forever but nothing ever gets done about it.

Plotting revenge on our writers (à la Douglas Robinson), making fun of their writing, complaining about the education system that produced them—these are doubtless of great therapeutic value. But in the end, they are futile: the text still has to be translated. So if we abandon the idea of our writers doing something about it, is there anything we can do at our end? If we turn from complaining to analyzing, can we develop techniques for dealing with poorly written texts? Better yet, can we avoid a hit-and-miss, case-by-case approach, and instead develop a systematic way of dealing with the problem?

In Part Two of this article, I offer my ’recipe’ for handling poorly written source texts, but in Part One I want to look more closely at the notion of bad writing, narrowing it down to those aspects which cause interpretation problems for translators.

The first thing worth noting is that there are degrees of bad writing. The meaning of a passage may be obscure on first reading, or when the passage is read quickly, but it becomes clear on a second, slower reading. Here is an example from a forestry text about jack pines in Quebec:

  • Nous avons pu observer que, dans certains cas, où un premier peuplement de pin gris s’était établi à la suite d’un incendie forestier et qu’une vingtaine d’années après le premier feu, il était survenu un léger feu de surface, probablement au printemps alors qu’apparemment les arbres étaient encore entourés de neige mais que celle-ci était disparue dans les ouvertures nombreuses du jeune peuplement, il s’était établi une nouvelle régénération de pin gris.

There is no real problem of interpretation here, though it may take a while to discover the structure of this sentence (nous avons pu observer queil s’était établi une régénération de pin grisdans certains cas où).

There are also several causes of bad writing. Certain problems arise during physical production of the text; others are due to the writer’s lack of language knowledge, and still others are caused by the writer’s failure to think about his or her readers during the composing process. Causes are discussed in section 2 below.

Finally, there are many different manifestations of bad writing: poorly organized argument, inconsistent level of language, inconsistent terminology, misplaced sentence focus, sins against ’correct usage’, ambiguous sentence structure, typographical errors, jargon, unexplained abbreviations, mixed metaphors and so on (and on and on). One reason for the length of the article you are now reading is that there are so many things that can go wrong when the source text is composed. There is simply no way of discussing the problem briefly.

Fortunately, not all manifestations of bad writing are relevant to our work as translators. Some aspects of bad writing will cause problems for source-language readers, editors or teachers; what we are interested in as translators is bad writing which causes a translation problem.

1. Three Preliminary Questions

1.1 Is the problem source-language interpretation or target-language composition?

Bad writing can create either problems in interpreting the text in the source language or problems in composing the translation in the target language.

A text full of acronyms, bureaucratic shorthand or inappropriate mixes of formal and informal language may be perfectly intelligible to an experienced translator, but if the translation is to be read by the general public, then the translator is going to have to do a great deal of the editing work which the original writer ought to have done.

Similarly, a text in a genre which the writer has not yet mastered (briefing notes for the minister, say) may be clear in meaning but wrongly structured, and this will create composing problems too: perhaps major reformatting will be needed, or sentences will have to be combined or reordered.

Another possibility is that the writer’s meaning may be clear from a diagram or photograph in the text, but the corresponding verbal description may be extremely unclear. The translator then has to compose new sentences based on the picture.

Here are some sample texts that would give rise to composition rather than interpretation problems:

Example 1

Many journalists respond to the time pressures of their work by larding their writing with buzzwords and clichés: Time is perilously fast-tracking towards the United Nations’ January 15 deadline, I read in a December 1990 news article about the Persian Gulf crisis. How can time fast-track? And how can it do so perilously? This will be deemed bad writing by many readers (and should have been by the editor), but there is no interpretation problem as long as the translator is familiar with current English and the events described. The meaning is clear.

Example 2

Sign in a convenience store: Sale of tobacco products is restricted to those 18 years of age and younger. This sign says the opposite of what is intended, but readers’ general knowledge will make the intent perfectly clear.

Example 3

Handwritten note attached to the exercise bicycles at the gym I attend: Not let sweat fall on. English is the manager’s second language, but his meaning is clear. This is a problem for a language teacher.

In this article, I will be concerned with bad writing only insofar as it creates problems of interpretation rather than composition. The difficult issues raised by Douglas Robinson in my epigraph—those of editing and re-composing while translating—would require a separate article.2

To interpret is to determine the point of the words on the page—how a particular sequence of words having certain dictionary meanings is connected to what lies outside it: other parts of the text, other texts, the real or imagined extratextual world. Three sorts of problem can arise:

  1. A passage has two or more possible interpretations, but it is not at all clear which is the intended one. Problem: choose one interpretation.
  2. A passage seems to be vaguely gesturing at a meaning, but no interpretation definite enough to serve as a basis for translation comes to mind. Problem: generate some interpretations from which to choose.
  3. The passage is not ambiguous or vague but it clashes with some other part of the text. The result may be perceived as inconsistency, nonsense or lack of logical flow. Problem: decide which of the two conflicting passages is wrong.

All three situations—but especially (2)—will elicit from the translator that all-too-familiar oh no! reaction. But before groaning inwardly (or aloud!) and setting about the often laborious task of interpretation, there is an important question you need to ask which may save you a lot of work:

1.2 Is interpretation necessary here?

It is generally recognized that the depth of understanding a translator needs is not as high as that required by certain other types of reader—for example, a student who has to pass an exam on the content of the text, or a colleague of the author who has to write a commentary on the text. On the other hand, as Pliny the Younger observed some 1900 years ago, quae legentem fefellissent, transferentem fugere non possunt: while students or commentators reading a text might skip over an obscure passage here and there, we poor transferentes generally have to come up with an interpretation of every last word.

However if a text is being translated for information only, failure to interpret this or that short passage may be acceptable, and the depth of understanding required will be lower than with texts being translated for publication.

Also important to bear in mind is that ambiguity of sentence structure or vocabulary may be deliberate, so that no decision need be made. And even when ambiguity is not intended, the translator can often reproduce it. Thus the syntax of dépenses reliées aux taxes et frais d’immatriculation suggests that immatriculation applies not just to the frais but to the taxes as well. Suppose, however, that the context makes the application to taxes doubtful (i.e. the text should have read … aux taxes ainsi qu’aux frais d’immatriculation). Solution: invert the order and write licensing costs and fees, which allows for both interpretations.

A similar point can be made about vagueness. Vagueness is an inherent property of all language, since speakers and writers need only be specific about matters relevant to their intent. Here is a vague sentence in a text about pay and benefits:

  • Employees were surprised to see their earliest retirement date.

Were the employees surprised that the date was mentioned at all? Were they surprised because they had never thought about their retirement before? Were they surprised that they couldn’t retire earlier, or the opposite—that they could retire so soon? The writer doesn’t say, and the answer isn’t implicit elsewhere in the text. But do we need to know what exactly was intended here in order to translate? If not, then there is no translation problem.

So before posing an interpretive question, be sure you need to know the answer. Once you have determined that interpretation is indeed necessary, a third question arises:

1.3 Is this a poorly written text or a difficult text?

Poorly written texts have to be separated from two kinds of difficult text—those which are intrinsically hard to understand, and those which the translator has difficulty reading because he or she lacks information which would be known by or available to readers of the source text.

It is important to remember that the plain clear style is not a goal in all kinds of writing. Aside from cases where the aim of obscurity is deception, examples of what might be called valuable obscurity abound in literature and law. A book of poetry or philosophy may be intrinsically hard to understand because the writer is struggling to introduce new modes of thought and feeling. A legal text may require several readings because the writer’s aim was not to lighten the reader’s task but to state the law, or the terms of a contract, in a completely unambiguous way.

In this article I will not be considering intrinsically difficult texts, but only texts in which we can assume the writer wanted to convey some relatively ordinary ideas in a clear way but failed to do so.

As for texts whose interpretation is difficult because the translator lacks information or knowledge available to source-text readers, obscurities can arise from:

  • unknown terminology, phraseology, acronyms or neologisms;
  • unknown concepts or ideologies;
  • unfamiliar genres, perhaps with unfamiliar grammar (e.g. telexes);
  • dependence on another text or on a picture;
    (e.g. comments on a proposal, where the proposal itself is not available; a verbal description of a complex physical object, where a photograph or diagram is not available).

This sort of difficult text requires research: calling the author or other experts, going to libraries, pleading for documentation and so forth. Calling the author is, of course, sometimes an option with bad writing as well, but it demands a great deal of tact, and may be impractical if the number of problematic passages is large.

It is very important to decide whether the problem with a text is difficulty or bad writing. Unfortunately, this is not always easy: what looks at first like technical writing may in fact be bad writing and vice versa. Then again, some passages may be simultaneously difficult and poorly written. A wrong assessment of the problem as bad writing rather than difficult writing can lead to disaster, for if the writing is bad, the translator usually has licence to edit and rewrite (that is, correct and recompose the text while translating it), but no such licence exists with texts that are difficult but reasonably well written.

Consider the expression winter severe weather in a meteorology text. This may appear to contain a language error, but if you mentally change it to severe winter weather and then translate it as if it were a descriptive phrase invented by the writer (say, temps hivernal rigoureux), the result would be wrong. Winter severe weather contrasts with summer severe weather, and severe weather is defined in terms of specific wind speeds, amounts of precipitation and so forth. There is a set French equivalent: temps violent d’été/d’hiver.

As a further example, here is a passage from a paper on the impact of telematics on lifestyles, originally read at a conference. It includes difficulty in the form of sociological jargon (bolded), one definite writing error (italicized) and one error of an uncertain nature (underlined: it is not clear to me whether the interpretation problem here is due to poor syntactic structure or to my lack of sociological knowledge).

Si la notion de style de vie permet la photographie des équilibres socio-culturels et donc de déceler les mutations culturelles en cours par superposition de sondages, elle oublie qu’une société ne peut être analysée que par addition de comportements individuels, et qu’il existe des groupes sociaux moteurs d’une dynamique qui échappe aux individus et leur dicte des valeurs.

The first part of the sentence means that one can see cultural changes by examining a chronological sequence of surveys, each showing the state of society at a particular point in time.

The italicized expression appears to assert that society can be analyzed only by adding together individual behaviours. However, this clashes with the rest of the sentence and with the flow of the argument as developed in the preceding paragraphs. The point would seem to be that society can not be so analyzed. Given the frequency of errors by French writers when they use the ne … que construction, we may mentally correct the French to ne peut pas être analysée par (simple) addition de …

The underlined passage has something to do with a group dynamics operating in society, a dynamics which transcends (operates beyond, or perhaps escapes from) the control of individuals. It is not clear whether social groups generate the dynamics, control it, or merely serve as a vehicle for it. And does the style de vie approach ignore the existence of just those social groups which constitute a driving force in society or does it ignore the existence of social groups in general, as opposed to individuals (i.e. should a comma be read in after sociaux)?

In the present article, I will not be considering texts which are difficult to interpret because the translator lacks knowledge. I will be considering only those cases where the obscurities have correctly been identified as being attributable to the writer of the source text.

2. Some Causes of Hard-to-Interpret Writing

In this section, I will try to answer the question "where do poorly written source texts come from?" The answer is of interest not so much because it provides a convenient categorization of the many diverse manifestations of bad writing, but because it can help pinpoint what exactly is causing an interpretation problem. For example, it may not be immediately obvious that the reason you are having difficulty interpreting a passage is that there is a typographical error resulting from inattention during physical production of the text. Perhaps the text reads 30 ou moins when it should read 30 au moins, or adopter les procédures when it should read adapter. But if you have mentally identified the text as one with physical production problems, that might point you in the right direction. You may be able to work backward from the erroneous text, through a hypothesized cause, to a correct version of the text.

2.1 Problems were created during physical production of the text.

If the text was originally produced on a word processor, the author may have erred during operations such as Cut & Paste or Delete. He or she may have forgotten to make appropriate changes after rearranging a sentence, or deleted too many or too few words. Another possibility is that the author may have incorrectly copied information from source materials.

Problems may also be due to a transcriber who erred when typing up what was originally a handwritten text. And then there are cases where the customer’s only copy of a text is in hard-to-read handwriting, or it is a faint photocopy, a printout that contains printer-generated errors, or an e-mail document in which all accented French letters have vanished.

Do not underestimate the possibility of physical production errors causing interpretation problems. In my experience, such cases are not at all infrequent.

2.2 The source text is in the author’s second language.

Suppose (as happened to me once) your client is in charge of the bilingual publication of a collection of scientific papers, and is requesting translation into English of a hydrology paper written in French by a native speaker of Serbo-Croatian. It is full of odd word combinations, peculiar syntactic structures and puzzling inter-sentence connections, probably reflecting the rhetorical habits prevalent in that language as well as overreliance on the equivalents found in Serbo-Croatian/French dictionaries.

Texts of this type call for consultation with someone who knows the author’s first language. If that is not possible, you may have to translate problem passages literally (that is, use the most common bilingual dictionary equivalents without regard to context) and leave much of the decision-making to the scientific editor. Or the editor may ask you to try a combination of summarizing and intelligent guesswork.

2.3 There are gaps in the writer’s knowledge of his/her own language (or language errors have not been edited out).

The French writer may not be fully aware of the faux-ami problem and may write si je passe l’examen, intending pass the exam, but the context is equally compatible with take the exam.

The French writer may never have mastered the rules for agreement of participles and may have written poubelles de rebuts vendus à notre client where the immediate context mentions sale of the bins themselves while the remoter context mentions sale of the scrap, and an ambiguous translation will not do.

There are of course many types of language-knowledge deficiency that can make interpretation difficult. I will not try to list them all here. However, one problem that is often overlooked is worth special mention, namely that everyone uses a few words of their own language in an idiosyncratic way. Consider these two sentences, the first from a news story about a statement by the Governor of the Bank of Canada, the second from the gardening column of a community newspaper, discussing ornamental grasses:

  • He denied the Bank was actively pursuing a high-dollar policy as a means of keeping inflation under wraps.
  • The varied colours and textures of their foliages and swaying flower spikes offer a colourful and restive scene which can rival any field of golden wheat or waving green oats.

The first writer seems to think that under wraps means under control, the second that restive means restful (unless restive is a misprint for festive!). Here is a French example, from a report on users of farm weather forecasts:

  • 75 % savent ce qu’indique l’indice d’assèchement, 88 % ne savent pas comment utiliser l’indice d’assèchement et 51 % ne savent pas ce qu’indique la probabilité de 5 mm [de précipitation]. Une des raisons de sous-utilisation de certains éléments est attribuable à la faible compréhension de l’utilité de ces éléments. En effet, presque la majorité des répondants ne sait pas comment se servir de l’indice d’assèchement et la moitié ne sait pas exactement ce qu’indique la probabilité de 5 mm.

How can 88% be almost the majority? The writer of this text consistently uses majorité where others would write totalité.

Here is a case, from a text on work standards, where it is not so obvious that a language error is causing the problem:

Asked if his association fears it will lead to uniformed security guards duplicating or cutting into the work of police officers, he says he doesn’t think so, because the standard actually complements the work of police and uniformed security guards.

The meaning is that the standard, if followed, would make the jobs of police and security guards complement each other. The obscurity arises because the writer wrongly thinks that the verb complement can be used in the same way as a verb like validate: to convey the meaning make the railway pass valid, you can write validate the railway pass but to convey the meaning make x and y complement [each other] you cannot write complement x and y. The French translator wrote La norme permet de mieux distinguer le travail…, which captures much of the meaning but does not make clear the complementarity of the two jobs under the standard.

In deciding whether language error is the problem, an important indicator is the number of clear cases of such errors elsewhere in the text. If there are many clear examples of defective language knowledge or failure to edit, you can feel more comfortable about silently correcting the author when interpreting. However, as will be indicated in Part Two of this article, great care must be exercised when making such corrections, even when they appear to be obvious.

2.4 The writer has not mastered a particular style or genre.

Writers may fill the text with the typical terms and phrasings of a style or genre which is new to them, in order to signal their membership in the group which uses that style. Frequently they will use the terms or phrases incorrectly (cf. the discussion of idiosyncratic usage in section 2.3), or use a technical term to express a non-technical meaning.

Writers of such texts are often newly appointed to their positions and may be experiencing insecurity. To compensate for this, they may try to impress readers by using ’ultracorrect’ language, and this too can create problems of interpretation. For example, an Anglophone attempting to avoid a split infinitive may create an ambiguous sentence such as He asked us clearly to underline the main points.

One particular problem for French-to-English translators is that most Francophone writers were taught at school not to repeat words but rather to use synonyms. Unfortunately, many never learned to avoid carrying this habit over into technical and scientific writing, where it can create enormous confusion:

Le papillon de P. quercicella est d’aspect brunâtre. Les ailes antérieures présentent une frange sombre et une zone médiane ombrée. … Le papillon de P. reflexella … est également brunâtre mais ne présente pas de zones foncées.

Do the three italicized modifiers have the same meaning? This passage is part of a detailed description in which several related species are distinguished, so small differences of shading could be important.

The comparison également … mais might suggest that foncé means the same as ombré here, but the fact that the second zones is plural suggests a reference to both the fringe and the medial area of the wings. If this is right, then foncé might cover both sombre and ombré, and could perhaps safely be rendered by dark. But does ombré mean the same as sombre or is it somewhat less foncé?

In the absence of high-quality colour photographs, what should the translator do? In Part Two of the article, I suggest a general principle for such cases.

2.5 The writer has forgotten that the intended readers are not part of his or her group.

Bureaucratese in texts for the public is perhaps the outstanding example of this problem. Writing which originates inside both public and private bureaucracies is noted for its excess verbiage, which leads readers to take certain expressions as adding to the meaning when in fact they are redundant. It is also noted for the opposite problem—its tendency to vagueness and inexplicitness (using passe-partout words; leaving implicit what needs explaining to an outside audience).

2.6 The writer is behaving as if the intended readers were in his or her immediate presence, and is writing as if speaking.

Here I am thinking of:

  • texts by semiliterate writers;
  • lapses into the spoken mode by literate writers;
  • unedited or semi-edited transcripts of speech, such as court or conference proceedings and parliamentary debates (not strictly speaking writing, but includable under this heading for present purposes).

Since the influence of speech on written texts is a factor not often considered in discussions of bad writing, I want to examine this matter in considerable detail. As translators, we are particularly apt to neglect the differences between speech and writing because we are professional literates who spend a large part of our waking hours in the world of writing—a very untypical human experience.

I will be looking only at lapses into the spoken mode by literate writers, though some of what I say may be applicable to transcripts and to semi-illiterate writing. Note that by speech, I refer to spontaneous face-to-face conversation, not telephone conversation or formal public speaking—announcing on radio or delivering a sermon. (Public speaking is often scripted—it is writing read aloud—and even when it is not, it tends to take the written language as its model.)

3. Speech-Influenced Writing

Speaking one’s first language is a natural ability acquired without specific instruction during infancy, but writing is an invention, and the ability to use it is achieved—often imperfectly—only after long years of guided practice. According to Alice Horning [Teaching Writing as a Second Language, Southern Illinois University Press, 1986], learning to write is like learning a second language. As I will be suggesting, our source-text authors frequently slip back into the habits of their first language—speech.

Now in everyday usage, we tend to refer to writing as if it were speech, using the verb say: what does her article say?, the author says that … Indeed, it is widely held that writing is best when it is a reflection of speech. The truth of the matter is that writing which was a pure reflection of speech would very often be incomprehensible. Anyone who has read an unedited transcript of a conversation or a speech in parliament will know this. On the other hand, there certainly are cases in which our source-text authors could have made themselves more understandable had they taken speech patterns as their model.

Thus problems in understanding English texts often arise from overuse of two kinds of noun-based syntactic structure that have developed in written English over the past few centuries, first in scientific and then in bureaucratic writing. Here is an example from Ernest Gowers’ Complete Plain Words [3rd edition, Penguin, 1985, p. 84]—a treasure trove of real examples of bad bureaucratic writing:

  • This compulsion is much regretted, but a large vehicle fleet operator restriction in mileage has now been made imperative in meeting the demand for petrol economy.

Is it a fleet of large vehicles or a large fleet of vehicles? And who is responsible for reducing mileage: the drivers by selecting different routes? the dispatchers by assigning vehicles that use less fuel? the managers of the fleet by reducing requests to use vehicles? Long sequences of noun modifiers (large vehicle fleet operator restriction) are extremely rare in speech: verbs are used instead, or prepositions are added to make relationships clear. If the writer had asked how would I say this to someone, the result might have been both unambiguous and easier to read:

  • We much regret having to do this but we have been obliged to greatly reduce the use of our fleet of large vehicles [or place restrictions on the operators of our large fleet of vehicles] in order to meet the demand that we economize on petrol.

Now consider the sentence: verification of the return of the samples is essential. Who is doing the verifying? Who is returning the samples? Or is the situation one in which the samples are seen as coming back automatically: the samples return rather than someone returns the samples, which focusses on the agent? Also, at what point in time does the verification occur: is it a check that the samples have returned to their point of origin already, or a check that whoever had them has sent them back, so that eventually they will return to the point of origin?

Had the writer used verbs, as in speech, these problems of interpretation would likely not have arisen. Finite verbs have the advantage over nouns of always showing tense and subject, so that we get answers to the questions when? and who?.

Granted, then, the value of imitating certain features of speech, it remains that speech and writing are fundamentally different modes of communication. The written mode is abstracted from any immediate situation; there is no direct encounter between writer and readers. Successful writers must compensate for two key absences in writing:

1. Speakers can clarify their meaning in response to a puzzled look, a question, or some remark by the listener which indicates misunderstanding. The single greatest speech-influenced error in writing is a simple failure to edit, because there is no immediate prompting from readers to do so. The writer fails to ask, or has never learned to ask, whether the readers will be able to recover the intended meaning from the words written. The non-presence of the readership during writing means that to be successful, writers have to imagine their readers and anticipate readers’ difficulties.

In a text on the need for air bags in cars to reduce the number of injuries during accidents, the authors mention les traumatismes parmi les plus coûteux (tête, colonne et tronc). This would probably evoke a puzzled look from a listener if it were speech. After all, the spine is part of the trunk, and more importantly, once you’ve subtracted the head and trunk, all that remains are the arms and legs as sites for less expensive injuries as well as the other expensive injuries implied by parmi. Perhaps parmi is empty verbiage. Perhaps the writers meant chest or abdomen rather than trunk. The parts of this expression just don’t fit together. It is inconsistent.

Another common feature of speech-influenced writing, also arising from the lesser need in speech to consider the listener’s ability to recover the intended meaning, is a problem I call apparent clarity. Here, the most obvious dictionary meaning of an expression makes perfect sense in context, yet is not what the writer intended. This is of course a simple failure to communicate on the writer’s part.

There is a way of detecting apparent clarity, but it is extremely time-consuming. In normal reading, we tend to stop interpreting as soon as we arrive at a plausible interpretation. But if, as translators, we wanted to avoid the apparent clarity problem, we would have to stop at each expression to consider whether the context will support interpretations other than the first plausible one we hit on. Consider this passage from a meteorology text about forecasting the polar low-pressure areas which bear severe storms:

  • Nous croyons qu’il est souhaitable de bien prévoir les dépressions polaires pour améliorer nos prévisions et nous croyons que nous pouvons y réussir à court terme.

The author wrote à court terme without thought to the reader’s likely first interpretation, namely bientôt. It made perfect sense in context to say that good storm forecasts would soon be possible, and that is how I translated it at first. I was not alerted to another possibility until I connected this passage with another one elsewhere in the text:

  • Il est donc possible de prévoir à court terme le développement de ces dépressions.

This suggests that the author’s real intent in the first passage (as indeed he confirmed during a telephone conversation) was that good short-term forecasts would be possible (i.e. forecasting a polar low a few hours before it forms).

Perhaps the author just assumed that meteorologists reading his text would automatically interpret à court terme as a reference to the forecasting period, even though he had constructed his sentence in such a way that, grammatically, the expression was modifying the verb réussir. Had I stopped and considered other interpretations, I might have elaborated the pronoun y and paraphrased the last part of the sentence as réussir à les prévoir à court terme, but there was no particular reason to do this.

Here is another case, from a job description, where careless word placement by the writer creates false clarity:

  • Administre les programmes de relations de travail et de relations humaines … en contrôlant les nominations faites pour une période déterminée par les gestionnaires délégués afin de …

A student translator wrongly rendered the italicized passage as follows: monitoring appointments made during a specific period by delegated managers. If the writer had placed faites immediately before par les gestionnaires, the translator would have been more inclined to investigate the expression nomination pour une période déterminée, a set phrase used to describe term appointments in the Canadian public service. In speech, such carefulness in word placement is unnecessary.

2. The second important absence in writing is the lack of what might be called embodiment. Speech is, quite literally, embodied. The speaker’s voice, gestures and stance play a vital role in conveying meaning. Anyone who has read a transcript will have found that speech can be almost incomprehensible when separated from any indication of stress and intonation, tone of voice, loudness, speed of utterance, facial expressions, hand and head gestures, body posture, meaningful silences, laughter or the spatial orientation of speaker and addressee with respect to each other and their surroundings.

Thus the reference of a pronoun might be made clear by tilting the head backward. The meaning of this gesture might be when I say one of those, I refer to a prominent object, relevant to our current conversation, which is behind me in your visual field, but which I don’t want anyone seated near us to hear me mention. Such a gesture would of course be absent in the written transcript, yet it could be necessary for identifying the object and thus achieving a correct translation.

Successful writing must compensate for the absence of gestures and other features that accompany words and syntactic structures in speech. Features that accompany written words—capitalization, commas, underlining, parentheses, point-form lists and so forth—will often be of value for this purpose (though they are not direct counterparts of the features that accompany speech). Here is a sentence that lacks compensation for a speech feature:

  • As these studies tend to show the form translation has taken in Canada, both on an institutional level and on the level of the actual practice of translation, is specific to our particular national context.

The lack of a comma after show—to compensate for the absence of the intonation pattern which in speech would indicate a clause boundary—leads readers to assume that form is the direct object of show, whereas in fact it is the subject of is specific. The writer may have been hearing the appropriate intonation in his head, but failed to transfer this effect to the reader, who is led up the garden path to the wrong syntactic structure. See section 5.2 in Part Two of the article for more examples of the missing comma problem.

Let us now look at four specific harmful effects of speech on writing.

3.1 Wrong expression in focus

In spoken English (and to a lesser degree French), stress often indicates which expression is in focus. In writing, focus is usually indicated by position, though occasionally italics or underlining are used. A speech-influenced writer will mentally stress the right expression while composing, but fail to notice that, given the syntactic structure, readers are likely to place the focus elsewhere, at least on first reading. This bad habit yields sentences such as the following (adapted from Gowers’ Complete Plain Words, p. 96):

  • His condition can only be alleviated by surgery.

Here the writer may be placing mental stress on alleviated, yielding the meaning can only be made more bearable, not corrected, but the first-time reader will most likely stress surgery, thus arriving at the other meaning of alleviatecan be corrected or partly eliminated, though only by surgery.

Here is a French example of poor focussing:

  • La biotechnologie est un ensemble de disciplines scientifiques et technologiques de pointe mais qui s’applique à de nombreux secteurs de l’activité économique dont certains sont très traditionnels.

From general knowledge, one might presume that the point being made here is that biotechnology is new but (mais) it has been applied in some very traditional sectors of the economy. Its application in non-traditional sectors hardly merits comment. Unfortunately the sentence does not reflect this. It appears to contrast disciplines de pointe and nombreux secteurs… rather than de pointe and [secteurs] traditionnels. If the intended meaning is indeed the one just suggested, then traditionnels should not be in a separate relative clause (which comes across as a speech-like afterthought whose sole function is to make it possible to introduce the word certains along with traditionnels).

3.2 Anacoluthon

Frequently in speech, we abandon one syntactic structure and start another, as in this extract from a conversation between two teachers reported by Milroy & Milroy in Authority in Language (Routledge 1985 p. 141; see chapters 3, 4 & 7 for discussion of the differences between writing and speech):

  • This is something I usually spend one lesson arranging what they want to talk about.

The two structures here are: this is something I usually spend one lesson [on] and I usually spend one lesson arranging what they (the students) want to talk about. Sentences of this sort sometimes appear in poorly edited writing, especially writing produced with the aid of word processing software. Consider these two sentences, the first from a newspaper article on an Ontario politician, the second from an article in Terminology Update:

  • Mr Silipo, successful as chairman of the legislature’s committee on Ontario in Confederation, got him noticed in the right places.
  • … it would be appropriate for computational terminology researchers would do well to investigate the potential usefulness of existing knowledge-engineering technology … ( vol.  24(2), 1991, p. 4)

As in speech, the user of a word processor has less need to plan ahead. He or she knows that correction is easy, but of course it is also easy to forget about it! In both of the above passages, the meaning seems fairly clear on second reading. Here is a case which is not so clear:

  • Critics say selection of William Teron lacks imagination at Queen’s Park. [Queen’s Park is the location of Ontario’s legislature.—BM]

Perhaps the writer originally had critics say Queen’s Park lacks imagination in selecting WT or selection of WT shows lack of imagination at Queen’s Park, and then revised to selection … lacks imagination but forgot to delete at Queen’s Park. However, there is also a possibility that the writer intended to convey more directly the idea that WT lacks imagination.

Also encountered are what might be called lexical analogues of anacoluthon—a tendency to mix two set word combinations, as in:

  • Meech Lake has beyond a question of a doubt enhanced our cynicism in parliament as an effective instrument of government. [Meech Lake was the location of a meeting where an ultimately unsuccessful proposal to amend Canada’s constitution was drafted.BM]

The expression enhanced our cynicism [about] has been combined with [reduced our faith] in, and beyond [any] question has been combined with beyond a [shadow] of a doubt.

Here is an example of anacoluthon from the published version of a speech by a government inspector of insurance companies:

  • Je reconnais que pour un assureur-vie de carrière, c’est un sujet [la déontologie] qui revêt une grande importance, surtout dans un monde en pleine évolution et dans lequel la concurrence est de plus en plus féroce, et où la nature des contrats de service qui lient certains d’entre vous à des compagnies qui exigent l’exclusivité de représentation, sans toutefois offrir toute une gamme de produits adaptés aux besoins du public, peut réduire vos moyens de manoeuvre, et aussi par des véhicules financiers modernes que les institutions financières offrent, lesquels véhicules risquent à plus ou moins long terme de supplanter une partie de la sécurité recherchée par vos clients en offrant des garanties de gains à court terme plus alléchantes que celles que vous leur proposez.

The passage that begins et aussi par has no syntactic relationship of any kind to what proceeds it, and the relationship in meaning is somewhat unclear as well.

This sentence also illustrates the speech habit of piling on one clause after another as new thoughts occur to the speaker, a habit which can create great confusion if transferred to writing. In his book Spoken and Written Language [Oxford University Press 1989], Michael Halliday claims—though this is controversial—that speech tends to use simple words and complicated syntactic structure, whereas writing tends to use complicated words and relatively simple syntax. Compare the exigencies of penury with the things you have to do if you’re poor. The former is lexically dense, packing the meaning of ten words into four. The latter is syntactically complex, with an if-clause inside a relative clause. The greater the syntactic complexity, of course, the greater the possibility of ambiguous sentences in which the reader will get lost.

3.3 Wrong word

One important feature of speech is that you have to keep talking. You can’t pause for half a minute if the right word doesn’t come to mind. As a result, speech is full of word choices that are slightly off target. If a writer is working quickly and fails to edit, the same thing will happen, as in this passage from a job description:

  • Administrateur, poste à l’étranger, dont les décisions et recommandations ont des répercussions sur environ 75 employés ainsi que sur l’apparence de l’ambassade.

The context does not immediately rule out the sense of physical appearance, as in on a repeint la maison pour lui donner une belle apparence. The person holding this position could be in charge of the embassy building, amongst other duties. However, further reflection leads to the conclusion that the intended meaning is image, a meaning related to, though not the same as, that found in sauver les apparences.

Here is an English example of an odd word choice. In this case, the writer’s intention is not so obvious:

  • Bank machines, photocopiers and central heating are a few examples from an almost infinite list of technologies and products that are domesticated and an indelible component of modern life.

What exactly did the writer mean by indelible? Sometimes it just means permanent (indelible ink), but more often it enters into word combinations that express a positive or negative attitude (indelible stain on his reputation, indelible memory of a loving father). The synonyms Roget lists are mostly of the non-neutral kind: inexpungeable, unforgettable and the like. So perhaps the writer is expressing a positive or negative attitude toward technology rather than simply observing a fact, but it is not clear.

3.4 Vagueness and Ellipsis

Speakers can often leave a concept unspecified where a writer should not. This is so for several reasons: misunderstandings have a good chance of being cleared up as the conversation proceeds (the listener either expresses puzzlement or makes a comment that indicates misunderstanding); speakers very often know their listeners and thus know what knowledge can be assumed and left implicit; finally, speaker and listener share an immediate situation to which the words may be referring.

The result is inappropriate vagueness (use of too general a word) or ellipsis (words are left out). Consider this sentence from the description of a secretary’s job at an employment insurance office:

  • Le poste exige du titulaire de fournir des explications limitées, tel l’état d’avancement des demandes de prestation.

In employment insurance texts, demande de prestation usually means benefit claim. But in this instance, the secretary is required not to explain progress on processing claims but rather to explain what progress has been made on secretarial services requested by claims officers. The writer was thinking demande de prestation de services but abbreviated it. This failure to specify the nature of the prestation opened up the possibility—indeed the likelihood—of an incorrect interpretation.

There are numerous ways of classifying interpretation problems. Classifying them in linguistic terms, we find three broad types: defects in the physical nature of the text (section 2.1), errors in grammar and vocabulary (sections 2.2 and 2.3), and problems in the actual writing work—in the composition of the text and its resulting structure (sections 2.4 to 2.6). Section 2 also used a classification in terms of cause, that is, in terms of the writer’s knowledge, psychology and activity (see the six subsection headings). Finally, a classification in terms of result would be possible, perhaps with a division of the problems into ambiguity, vagueness and inconsistency.


Classification can help us become more aware of the problems we are up against, but as Part Two of the article will make clear, a three-part or six-part classification of the problem cannot lead to a simple three-part or six-part solution. Not only does each category in any classification cover a vast number of specific things that can go wrong during the preparation of the text, but also, and more importantly, it can often be difficult to determine which of these things is causing the interpretation problem.

Part Two: Introduction

In the first part of this article, the concept of poor writing was analyzed and some of its causes were examined. In this second part, I want to suggest a procedure for interpreting such writing.

The challenge in dealing with poorly written source texts is that of identifying the problem in the first place and then thinking of all the possible ways of solving it. What one wants to avoid is staring at the text in horror, or reading it over and over, hoping against hope that the meaning will suddenly come through in an inspired flash.

Actually, the more often you read the problematic passage, the more likely you are to become fixated on one possible identification of the problem, quite possibly the wrong one. I can remember once staring at a text containing the expression poubelles de rebuts ménagés and asking myself which meaning of the verb ménager might be close to the intended meaning. Did it have something to do with economizing on garbage or saving it up for a rainy day? I had unthinkingly decided that the writer had chosen a word that was not quite the right one, and begun to search for related words.

What I ought to have done, of course, was ask myself the question: is there a typographical or spelling error here? The correct reading—rebuts ménagers—would then have come to me immediately. This is the sort of thing that native speakers of the source language tend to spot easily, but others can see it too if only they ask the right question. A true professional, after all, is not someone who has all the answers, but someone who knows how to ask potentially fruitful questions.

The procedure I suggest for interpreting a poorly written text is to systematically ask a list of questions about each problematic passage. Checklist One (section 4) offers an ordered set of questions for identifying the type of problem. If it’s clear what the problem is, then proceed directly to Checklist Two (section 5), a set of questions for generating interpretations of vague passages or choosing between interpretations of ambiguous passages. Checklist Three (section 6) points to solutions for drafting the translation if you have tried everything and the meaning is still obscure.

Before proceeding, I should make clear the status of my ’recipe’ for dealing with poorly written source texts. I believe that experienced translators who are good at what they do operate in two different modes, as do successful practitioners in other fields, whether chess, plumbing, dentistry or hockey. There is the normal operating mode and there is what I like to call the ’bump’ mode, when things go wrong.

In normal mode, people who have mastered some skill simply ’see’, instantly, how to proceed. In bump mode, however, principles have to be applied. Translators who are good at getting over the bumps in poorly written source texts do so, I think, by using mental checklists of principles, though not necessarily in a fully conscious way. Only if the bump is a very bad one does it become necessary to consciously ask questions: "I need to know the true intended syntactic structure of this expression. Specifically, which noun does this adjective modify? Did the writer mean poubelles vendues or rebuts vendus in poubelles de rebuts vendus à nos clients?"

Thus, the checklists I offer here may be read as a hypothesis about the mental procedure followed by a translator who is successfully dealing with a bump that may not be serious enough to warrant conscious reflection. The lists can, of course, also be used consciously as a way of improving your ability to handle poorly written source texts.

If you have access to hypertext3 software, you should be able to create an automated version of the checklists. This might take the form of a Help window that would display a variety of screens joined by hypertext links. The links would let you move rapidly through various sequences of checklist questions.

A word of general advice—always remember that no poorly written text is all bad. The bad writing is at specific locations, which you can tackle one at a time. You may be able to reduce an obscurity at one location without eliminating it, thus improving your final translation. Don’t stop too long on any one problem. You can come back to it later, and by that time, other portions of the text may have clarified the matter.

Dealing with a badly written text is not a matter of finding some overall solution but of finding individual solutions (sometimes partial) to individual problems. If you do not take this piecemeal approach, you run the risk of inwardly giving up, and then lapsing into a kind of semi-automated ’literal’ translation, in which you are not basing yourself on any degree of understanding at all, but simply trying to move directly from the wording in the source language to the wording in the target language.

4. Checklist One: Identifying the Problem

This list is a flow chart. Start at the top left and work down through the questions in the leftmost column. Numbers in brackets refer to the sections of Part One of the article where a topic is discussed or an example given.

Questions 1 to 3 distinguish problems that are not the focus of this article (see Part One, sections 1.1 to 1.3, for limited discussion).

If your answer to Question 4, 5 or 6 is yes, then move to the second column, which goes into more detail (but is not intended to be exhaustive).

The alphanumeric symbols in the third column suggest relationships to questions in Checklist Two, which will be discussed in section 5. The fourth column lists causes of bad writing. This can help identify the problem: if the text is a transcript, then the possibility of a physical production error is very high. Note that an item in the fourth column does not correspond solely to the item on the same line in the second column: the six causes beginning word processing error, for example, apply as a group to the physical production errors listed in the second column.

Thus if you have answered yes to Question 5 (meaning that you think the problem could well be a language error), and then used the second column to identify it more specifically as a possible case of idiosyncratic usage, then you go to Question 8 on Checklist Two, as indicated in the third column. If the source text happens to be written in the writer’s second language (fourth column) that will greatly increase the chance that the problem is indeed a language error.

You will probably find that each individual phenomenon mentioned on the checklist is a familiar one. The point of using the list is to make sure that you have thought of every possibility, of which there are a very large number indeed. For example, it may not be obvious that a sentence has a second possible syntactic interpretation; in such a case you would need to remember to ask yourself whether your problem is being caused by ambiguity.

Checklist One: Identifying the Problem
Question Answers and possible next steps Possible explanations


If No: it’s an interpretation problem. Go to next question

If Yes: improve language and composition (e.g. eliminate jargon, combine and reorder sentences or clauses,  translate from picture, reformat) and/or adapt (eg change level of lg, give explanatory translation)



If Yes, go to next question

If No: translate using achieved degree of understanding, possibly drawing on Checklist Three
  • deliberately vague or ambiguous
  • translation for information only
  • further understanding might be needed for other purposes, but not for translation


If No, go to next question.

If Yes: research

  • translator lacks knowledge; text intrinsically hard


If No, go to next question.

If Yes:
typographical error? Go to B5,6
punctuation problem? Go to B4
layout problem? Go to B4
too many words? Go to B7
words missing? Go to B7
correctly read if illegible?  Go to B6

  • word processing error
  • copying error by author
  • error by transcriber
  • bad handwriting
  • poor photocopy
  • computer printout error

(error in grammar or vocabulary)
(2.2 - 2.3)

If No, go to next question.

If Yes:
idiosyncratic usage? Go to C8
agreement error? Go to B6,4
misleading cognate? Go to C8

  • Source language is writer’s 2nd language (2.2)
  • writer lacks knowledge of 1st language (2.3)

(2.4 - 3.4)

If No: Can a source language colleague see the type of problem, or rephrase, or give meaning? If still No, go to Checklist Three or ask writer.

If Yes:
inconsistency? (3.0) Go to C10
ambiguity? (3.0) Go to D11, 13
wrong focus? (3.1) Go to B4
anacoluthon? (3.2) Go to B4
wrong word? (3.3) Go to C8
vagueness? (3.4) Go to C8
ellipsis? (3.4) Go to C8
unclear connector? Go to D12
is word x redundant? Go to C9
are words x and y synonyms? Go to C10

  • writing in an unfamiliar genre or style (2.4)
  • forgetting readers are not part of his/her group (2.5)
  • writing as if speaking (2.6)
  • not thinking of how readers will interpret (3.0)

5. Checklist Two: Generating Possible Interpretations

This is a list of sixteen questions to help lead you to an interpretation of a poorly written passage.

Group A consists of three preliminary questions that should always be asked first (not necessarily in the order given).

Group B contains four unordered questions related to physical production problems. These should be asked second, in order to be sure that you are considering the right linguistic forms, the ones the writer had in mind.

Groups C to E contain questions that point away from the dictionary meanings of the specific words you are interpreting and toward increasingly remote aspects of context.

Checklist Two: Generating Possible Interpretations

A1. Is the meaning revealed by trying to translate?

A2. Is the idea repeated later (or earlier) in the text?

A3. Have I taken the wording of the text seriously, assuming the author meant what he/she has written. Is it really bad writing?

B4. Is the meaning revealed by reading the passage aloud, with various intonations (various comma placements)?

B5. Is a phonetically similar word the right one?

B6. Is an orthographically related word the right one?

B7. Is the meaning revealed by rereading the passage, omitting or adding words?

C8. Does a sense-related expression reveal the meaning?

  • a more general word?
  • a more specific word?
  • a near-synonym?
  • the negative of an antonym?
  • an explanatory paraphrase?
  • a summary based on keywords?

C9. Can I assume that expression x does not add new meaning but is redundant?

C10. Can I assume that expression y has the same meaning as expression x?

D11. What meaning is suggested by the rest of the sentence?

D12. Is the meaning revealed by considering the function of the sentence within the paragraph, as opposed to asking what it means?

D13. What meaning is suggested by:

  • the paragraph?
  • the section or chapter?
  • accompanying diagrams or illustrations?
  • the text as a whole?

E14. What meaning is suggested by my knowledge of the situation that gave rise to the text?

E15. What meaning is suggested by a related text?

E16. What meaning is suggested by my knowledge of the subject matter?

5.1 Group A

Question 1

You may find Question 1 puzzling. How can I translate, you may wonder, before I’ve arrived at an interpretation? Surely the result can only be a bad literal translation.

I would suggest, on the contrary, that the way to translate at least some problem passages, if a few moments of interpretive work yield nothing, is to start translating.

Many translators try to fully understand their text, badly written or not, before they begin composing the translation. They conceive of writing, whether translating or ’original’ writing, as a process of setting down thoughts that are already in the head. I have found, on the contrary, that what I am going to say about my subject often comes into my head while I am composing. I proceed by putting words down on the page, not by staring into space, or at the source text.

As the 19th century German writer Heinrich von Kleist put it, in a wonderful short essay that has been translated under the title On the Gradual Fabrication of Thoughts While Speaking:

  • As they say in France, l’appétit vient en mangeant and from our own experience we might in parody assert, l’idée vient en parlant… Often, while at my desk working, I search for the best approach to some involved problem. I usually stare into my lamp … while striving with utmost concentration to enlighten myself. … And the remarkable thing is that if I talk about it with my sister, who is working in the same room, I suddenly realize things that hours of brooding had perhaps been unable to yield.

When I write something like the article you are now reading, I simply start with some general themes in mind and perhaps a page or two of scattered jottings, certainly nothing remotely resembling an outline. I use the writing process as a method of thinking about what I will say, not just as a method of communicating.

The same is true of my translation work: I use drafting as a method of thinking about what the source text means. I start with just a few general ideas about what is in the text, some of them perhaps gleaned from preliminary research. I work out most of the problems of interpretation while drafting the translation, and I find that many points which were obscure during reading cease to be obscure when I actually try to set down a translation. If the solution does not come immediately, I leave a blank or make a guess, inserting a highlighted question mark.

The first-understand-the-text-fully method may work with reasonably well written texts. But I would suggest that with badly written ones, it may be a recipe for despair. Instead, just start composing: let your fingers do the thinking.

(Different individuals, of course, find different writing strategies congenial; the think-by-writing method may not work for you. Daniel Chandler calls it the oil-painter approach to writing—as opposed to the architect, bricklayer and water-colourist approaches—in his article Writing strategies and writers’ tools, English Today, April 1993, pp. 32-38.)

Question 2

This question is based on the idea that all may be revealed in the fullness of time. That is, if you wait a few pages, or go back a few pages, you may find that the idea is repeated but with improved wording. Even if you are one of those translators who reads the source text quite thoroughly before beginning to translate, you may have forgotten something you read on page 20 by the time you get around to translating page 10.

Consider this passage from a list of problems which developing countries experience in gathering meteorological data:

  • difficultés de maintenir des sites d’observation avec des enregistrements homogènes avec le développement

How does avec le développement fit in? I found this passage totally obscure. The author of the report was a native speaker of Arabic, and I had found other odd uses of the preposition avec in the text. I left a blank and was relieved to find, a few pages later, the following light-shedding sentence:

  • Avec le développement économique et social, il devient très difficile de maintenir des sites d’observation en fonctionnement, de les préserver de toute altération et surtout de maintenir des séries homogènes.

The point would seem to be that it’s hard to maintain observation sites and keep uniform data series in a developing country.

If my source text had been on diskette, I could have used the Search function of my word processing program to look for other instances of avec or développement in the text.

Do not confuse Question 2 with Question 13 in Group D. Question 13 does not ask whether some other passage repeats a specific idea with improved wording; it asks whether some other passage provides additional information that might support a certain interpretation of the problem passage.

Question 3

This is perhaps the single most important question in the list. Many translators are in the habit—the very bad habit, I would say—of ignoring the actual wording a writer has used if it doesn’t immediately make sense, and instead following the context-is-everything principle: judging from context, my writer must have meant such-and-such. The temptation is especially great with poorly written texts, since once you have identified the text as poorly written, you may feel authorized to disregard the writer’s word choices.

The danger here is that of reading in something which makes sense to you. But the writer may have meant exactly what he or she wrote. Perhaps some dictionary sense of a word other than the most common one was intended, or a sense other than the one that first came to mind (your mind, that is). Or perhaps some other interpretation of the sentence’s syntactic structure will yield the right meaning.

What I am suggesting, quite seriously, is that even when you come across something like rebuts ménagés or keeping inflation under wraps, you should not immediately assume an error. Instead, you should first ask whether some meaning of the verb ménager is the right one, or, in the second example, whether something is indeed being kept secret. Only if nothing comes to mind should you proceed to substitute the nearest set combination of words (rebuts ménagers) or commonplace idea (keep inflation under control).

The importance of this approach is illustrated by the following passage from a text about assertiveness training for prison inmates:

  • Motifs sous-jacents à l’absence d’affirmation de soi: … le sujet, même s’il en possède l’habileté [de s’affirmer], ne l’utilise pas parce que le renforcement disponible pour l’émission d’un tel comportement recèle une valeur neutre ou négative. Prenons pour illustrer l’exemple classique du détenu qui préfère adopter un comportement inadéquat par crainte qu’un comportement adéquat lui mérite des félicitations de la part d’un administrateur.

Here is the translation of the last sentence submitted by a freelance:

  • A classic example is the inmate who prefers to adopt an inadequate type of behaviour for fear that adequate behaviour will result in his being criticized by a prison official.

Here the translator seems to have assumed that the writer got confused and wrote the opposite of what was intended. Why, after all, would anyone fear being congratulated? The translation does make sense: the prisoner is afraid that if he is assertive (i.e. if he displays suitable behaviour), he will be criticized by a prison official. But the French text as it stands makes perfect sense if the words are taken with their ordinary dictionary meanings: the prisoner would not want to be seen as the warden’s pet by other inmates, and therefore fears being congratulated.

Notice how this case illustrates the falsity of the view that meaning is determined by context. The meaning of any given expression is only partly determined by context. In part, it is determined by the dictionary meaning of the expression the writer has selected—obvious perhaps, but nonetheless frequently forgotten.

The first question to ask in translating any expression in any text is this: is there not some dictionary meaning of the problem word which makes sense here given the whole context? The difficulty, of course, lies in making sure you have not missed some aspect of the context. And context, unfortunately, as we all know, has a great many parts: the immediate verbal surroundings of the expression, the sentence, paragraph and section, the logic of the text as a whole, related texts, the real-life situation within which the text is embedded (e.g. the actual committee meeting, the minutes of which you are translating), the subject matter, and the general cultural background. See the introduction to Groups D and E on the order in which to consider the various elements of context.

When considering dictionary meanings, do not forget set expressions. It is easy to mistake these for inventions by the author. A few years ago, I mistakenly took grande pêche, in a fisheries text, to be an odd way of saying grande prise. Big catch was plausible in context, but as Le Robert or Harrap’s would have told me, grande pêche is a set expression meaning deep-sea fishing and by implication lengthy expedition—a fishing trip of several weeks or months, as opposed to a day trip in the inshore fishery.

Do not, then, abandon too hastily the out-of-context meanings (literal meanings, dictionary meanings) of the words before you—either the common ones or the not-so-common ones. If you are sure that none of them fits, try to find a meaning that involves the least possible modification of one of the dictionary meanings (see Question 8 below). Do not make grand leaps to meanings that are only distantly related to the dictionary meanings, just because the result will be plausible. You cannot wholly avoid such leaps, but you can make an effort to minimize them.

This requires banishing from your mind any notion, however tempting, that the author doesn’t really know what he means. While people often open their mouths and start speaking without much thought about what they want to say, writing by its nature provokes a certain degree of thoughtfulness. This may be less true now, with the advent of electronic writing, than it was in the days of manual typewriters or pen and paper. Nevertheless, as a general rule, confused writing does not imply confused thinking. There was some mental path that led the writer to the wording which appears on the page, if only you can recover it. Poor wording is not a signal to substitute your own ideas.

Two final points to conclude. First, note that the issue raised by Question 3 has to do with the interpretation of the source text, not the literalness or freeness of the translation. To avoid mistranslation, you should begin with a literal interpretation of the source text; how freely or literally you translate, once you have correctly determined the meaning, is a wholly unrelated matter.

Second, alongside passages which you may falsely identify as badly written, there may also be passages which you falsely identify as well written. The obvious dictionary meaning makes perfect sense in context, but it is in fact not the intended sense. This is the problem of apparent clarity discussed in section 3.0 of Part One.

5.2 Group B

Question 4

Don’t forget the possibility that a comma may have been omitted. Consider the following passage:

  • La biotechnologie … permet le développement des biocapteurs pour détecter les toxicités dans l’air et l’eau et par les lits bactériens immobilisés, le traitement des eaux et des effluents.

How does par les lits fit in here: permet le développement par les lits? détecter les toxicités par les lits? And how is le traitement … related to the rest of the sentence? Suppose you don’t know much about the subject matter (the problem here is difficulty as well as physical production). If you try reading the sentence with various intonations (that is, pausing first at one point, then at another), you will eventually arrive at the possibility that there should be a comma before par les lits …. The meaning is: permet … le traitement des eaux … par les lits bactériens. You could then guess that lits bactériens is a water treatment method, and do the appropriate research to confirm this solution.

Here is a case where rereading with various comma placements can resolve an apparent contradiction:

At a news conference today in San Francisco, IBM and Apple said they will disclose further details about their plans for linking computers, creating new software and advancing computer chip technology. The news conference will be held at the Fairmont Hotel …

In the first sentence, it sounds as if the news conference has already occurred; in the second sentence, this is contradicted. If the first sentence is read with a comma after said, the contradiction is resolved. The meaning would seem to be: IBM and Apple said that at a news conference to be held today in San Francisco, they would disclose further details …

Questions 5 and 6

The right word may be one which sounds similar to the word on the page (Question 5). These are cases where the transcriber misheard tape-recorded speech or was unable to interpret it. Alternatively (Question 6), the author’s or keyboarder’s fingers may have slipped while entering the source text, or you the translator may have misread bad handwriting or a faint photocopy. In both cases, you can sometimes get help from your word-processing software:

  • Entering a word in the Spell/Look Up option of WordPerfect 5.1 will give you a listing of words phonetically similar to the one you enter. Thus if a phonetic statement makes no sense to you, the Look Up option will suggest fanatic and phenetic. Perhaps the original speaker mentioned a fanatic statement. (Unfortunately, the corresponding command in the French version of WordPerfect does not yield such a listing.)
  • If you can make out some of the letters in a word fairly clearly, but not others, use the Wildcard option in your word processing program. With WordPerfect 5.1, for example, entering tr??e in Spell/Look Up will get you a list of 16 five-letter English words or 9 five-letter French words that start with tr- and end with -e.

Here is an interesting case, from a job description, which illustrates an interaction among the issues raised in Questions 2, 3 and 6:

  • Établir la liste de radiation des amendes dans les cas de décès lorsque l’accusé a quitté le pays ou que l’amende est non payée après plus de 36 mois et recommander la rédaction de ces amendes.

Does the person holding this job recommend that fine notices be written up? This interpretation can’t be entirely ruled out, and Question 3 requires us to take it seriously. Question 2 suggests that we look at other parts of the text to see whether the idea is repeated with better wording. Examination of the first part of this very sentence suggests that rédaction should read radiation. This is most likely a transcriber error.

Question 7

If the writer failed to edit a sentence after Pasting or Inserting material with the word processor, rereading the sentence with various omissions may make the meaning clear (see examples in section 3.2 of Part One). Alternatively, you might try to make additions, though this is trickier since it calls for reading in from context. Consider this passage from an Ann Landers column:

Dear Ann: I am the woman who found herself competing with her fiancé’s dog. I was once in a similar situation and think I can help her.

The second sentence is not compatible with the first. The writer cannot be both the woman competing with the dog and also the person in a similar situation. Since we know that people write to Ann Landers in response to previously published letters from others, we may assume that words got deleted at some stage during physical production of the newspaper, and correct to I am responding to the woman who …

5.3 Group C

Question 8

Perhaps a near-synonym or a word more general than the one before you will reveal the intended meaning. If the problem is vagueness, a more specific word might make things clear. Thesauruses will give you a list of near-synonyms, sometimes mixed with more specific or more general words. No commercially available English or French wordbook, to my knowledge, systematically lists more general and more specific words in separate subsections of each entry. The only exception is concrete nouns: picture dictionaries will list the subtypes of a type (under sail or mast you could find topgallant), and regular dictionaries will give the more general word if you look up the more specific one (under topgallant, you would find mast or sail as part of the definition or in an example).

When considering extended meanings of a word, don’t forget to try meanings of the cognate word in other languages. If the author is bilingual, or some other language is known to be affecting the source language, this could be the key to your problem.

Another trick is to pick a common target-language equivalent of the problem word and then substitute the negative of an antonym. Suppose communication latérale makes no sense. From lateral, you can go to sideways, then horizontal, then not vertical, then not top-down. Perhaps the author is talking about the need to avoid top-down communication.4 Or again, suppose approche logique has you baffled. You could move from logical to not illogical, then to not senseless and finally to makes sense. The author may be talking about an approach that makes sense.

I was once confused about the statement, in a government publication, that the Senate is not a confidence chamber. The expression confidence chamber brought to mind only confidence interval (from statistics) and confidence man (from criminology). One way I could have arrived at the correct interpretation would have been to think of the negative expression non-confidence chamber, which might have brought to mind non-confidence motion. The French translation of the document made the matter perfectly clear: … n’est pas une chambre dotée d’un vote de confiance.

Computerized thesauruses that list synonyms and antonyms of a word, and then let you move instantly to synonyms of those synonyms and antonyms, can be extremely helpful since you can scan a wide range of possibilities in a few seconds. It’s a mistake to think that synonym lists are useful only for style editing or for jogging your memory while drafting the translation (that is, when the right target-language word is on the tip of your tongue). Thesauruses—of both the source and target languages—can also help you think about the meaning of the source text.

The last two items in the listing under Question 8 refer to operations on sentences rather than words. Interpretation can sometimes be eased if you unpack an ambiguous expression or passage into a paraphrase. Consider the example discussed in section 3.0: verification of the return of the samples is essential. This could be turned into the sequence someone must verify the return plus someone will bring/send back the samples or find out whether the samples have come back. Each of these can then be considered in the light of the broader context.

If expansion to a paraphrase will help in some cases, in others the solution is to contract a verbose sentence into a summary using keywords. Consider the following passage from a court transcript (originally an oral text, but the principle applies equally well to writing, especially writing affected by speech). Defence counsel is making arguments on sentencing following a guilty plea on a charge of murder:

  • Alors compte tenu Votre Seigneurie que, l’individu évidemment, je ne parle pas d’un individu qui ne serait pas à ses premiers antécédents en semblable matière, et qui à ce moment-là, comme mon confrère vous l’a mentionné dans le cas de P…, que vraiment l’individu a tué délibérément, Votre Seigneurie, de sang-froid, je pense que là il y a une marge, ce n’est vraiment pas le cas.

This is really just an expansion of a simple idea: mon client n’est pas un individu qui aurait tué de sang-froid plusieurs fois. So the passage can be interpreted as follows: I am not speaking, Your Lordship, about an individual who has done such things before and, as in the P… case, killed deliberately in cold blood. This is really a different situation. (This wording might also be useable as the actual translation, as suggested in point 2 of Checklist Three below.)

Questions 9 and 10

Two common features of writing are redundant wording and the use of different expressions to mean the same thing. The main hazard here is that of automatically assuming such redundancy or synonymy. Always begin instead by assuming that each and every word is contributing some new meaning, and that different words have different meanings. Thus, in the scientific text discussed in section 2.4 of Part One, begin by assuming that sombre, ombrée and foncées have different meanings.

5.4 Groups D and E

If dictionary meanings (Question 3), formally related words (Questions 4-6) and words or sentences related in meaning (Question 8) don’t help, you will have to appeal to the accompanying text or to your knowledge of matters external to the text. But never hesitate to go back to questions in the earlier groups, even Group A.

The questions in groups D and E should be asked in the order shown. In other words, the meaning suggested by the surrounding sentences takes precedence over the meaning suggested by more remote parts of the text if the two interpretations conflict (Group D). Similarly, the meaning suggested by some other part of the text takes precedence over the meaning suggested by related texts or by your knowledge of the subject matter (Group E).

Why does text-internal evidence take precedence over evidence from outside? Because your knowledge of the subject matter is derived from previous reading, or from personal experience of earlier realities. However, the author is not necessarily referring to the same realities earlier texts referred to, or the realities with which you are familiar. Every text should be assumed to be leading the reader toward its own unique meaning.

The last item in Question 13 refers to the text as a whole. By this I mean that the overall pattern of meaning in a text—the topic and the overall structure and drift of the argument—will of course serve as a general background to interpretation. An obvious example: when considering the basic dictionary meanings of words in their sentential context, some possibilities will be ruled out by the topic of the text: nuage meaning cloud will be ruled out in a statistics text, where the basic meaning will be cluster (of points on a graph)—unless of course it’s a statistical study of clouds!

Similarly, you may be able to rule out an interpretation of a problem passage on the ground that it does not fit into the flow of the argument. But be careful: if the text is poorly written, the flow may not be very logical in the first place. Alternatively, you may have misperceived the drift of the argument (cf the inmate assertiveness text discussed under Question 3), or failed to notice that the problem passage is part of an ’aside’—a separate argument within the larger argument.

Question 12 also concerns the flow of the text, but this time at paragraph level. It asks you to consider the function of a sentence within a paragraph. The function may be to give the next event in a narrative, to give an example of a generalization, to qualify or clarify the preceding statement, to define a term, to summarize, and so forth. Sometimes a sentence won’t make any sense if you have (unconsciously) decided that its function is x, when in reality its function is y. Thus a sentence may appear to be providing information about the real-world referent of a word, but actually it is giving a definition of the word. Or what looks like new information is really a restatement in different words of what has just been stated.

Finally, newer translators often overlook the clarifying potential of diagrams, illustrations, tables and the like (the third item in Question 13). In a text on the relationship between highway accident injuries and non-use of seat belts, I couldn’t figure out the unexplained acronym EA until I noticed, on an accompanying table, that one column was headed Ens.Acc. Each number appearing under this abbreviation was the sum of the figures to the left of it in the same row. Only then did it dawn on me that Ens. was short for ensemble—the total number of accidents.

6. Checklist Three: Avoiding Interpretation Decisions

If nothing works, the question then arises whether you can somehow avoid interpreting an expression whose meaning, apparently, is destined to remain forever obscure. The solutions on Checklist Three are ones you will already have applied if your answer to Question 2 on Checklist One was no: if the text is for casual information only, there is no point in going through laborious interpretation procedures. But if you have gone through them, to no avail, then it’s time to reconsider the ’avoidance’ option.

Checklist Three: Avoiding Interpretation Decisions

  1. Can I avoid deciding among alternative interpretations by signalling the problem to the reader:
    • use alternatives in the text?
    • use question marks in the text?
    • use footnotes with alternatives, literal renderings, or other forms of explanation?
  2. Can I avoid deciding by omitting the problem expression:
    • leave out an idea?
    • summarize long, badly structured, redundancy-filled sentences starting from keywords?
    • summarize a quotation using indirect discourse?
    • refer the reader to a table or picture and omit the obscure verbal description of what the table or picture contains?
  3. Can I avoid deciding by fudging:
    • use an ambiguous syntactic structure?
    • use a vague word?
    • use a layout or punctuation device that leaves the problem unresolved?
  4. Can I use indirect discourse for the whole translation?

Comments on Checklist Three

Question 2, first item

Often you can leave an idea implicit and the knowledgeable reader will be able to recover it. Omission is also possible when the writer has included a parenthetical list of three or more examples to illustrate a point. If one of them is obscure, just leave it out.

Question 3, second item

If you can’t decide which of several possible meanings is the intended one (cf the passage with the word indelible in section 3.2 of Part One), and an ambiguous word or phrase is not available, select a more general word that fits the context. This technique also applies in cases where it is not even clear what the possibilities are, as in:

  • The union’s show of force on Monday did a lot for the strikers’ pride. The numbers surprised even the organizers and it sent a very strong message to the government that its workforce was becoming radicalized, with poignant implications for future relationships in the workplace.

None of the dictionary definitions of poignant really fits here, and the broader context does not help. The writer may originally have had some other word in mind (strong?, deplorable?) but decided to substitute a synonym from a thesaurus, without however having a very precise knowledge of the word selected. There may be little choice but to translate as if some suitable general word (e.g. significant) were present. Generally speaking, in cases where a footnote would be inappropriate, it is better to lose meaning through vagueness than to get the wrong meaning through guesswork.

Question 3, third item

Sometimes if you cannot see the connection between two sentences or two parts of a compound sentence, you can simply start a new sentence or new paragraph with no connecting expression. Consider the text about life insurers discussed in section 3.2. The solution to the obscure connector et aussi par is to start a new sentence or paragraph: Financial institutions also offer modern …

Question 4

If translating a semi-illiterate letter from a Mr. X, use the form  Mr. X writes that … This allows extensive omission, possibly to the point of giving only the gist of the letter. It also circumvents the problems which arise if you try to imitate bad writing or—the opposite strategy—you translate as if the author were well educated.


In an ideal world, either the schools would produce good writers or else large organizations would impose certain writing standards on all their documents and hire the editors needed to ensure compliance with these standards. In the actual world, schools produce too many mediocre writers and most documents are written by employees whose principal duties have nothing to do with writing, who receive no training in writing or editing, and who often do not enjoy writing. There is therefore no solution but to include in every translator’s job description: ability to interpret poorly written texts.

Applying the Method

There being no space here to reproduce long passages from texts, the following demonstrations of the method—in the form of ’exercises’—cannot be very realistic. The important thing is the mental process illustrated in the answers to exercises 1 and 2 (for exercises 3 and 4 I have simply listed the problems). If you want to try the exercises, assume that research on the topic would not help with interpretation. If you see the answer right away, ask yourself what procedure you might have used to arrive at it had you not seen it immediately. Alphanumeric symbols in brackets refer to the questions of Checklist Two.

Exercise 1

This is an instruction in a manual for training tobacco excise tax inspectors.

  • s’assurer que les notes de crédit ont été octroyées au client dans le but d’assurer le retour des marchandises

The customers are retailers returning tobacco products to the manufacturer. What does the instruction require the inspector to do? Where did the writer go wrong?

Answer. The first meaning that occurs to you is odd: why would the manufacturer give out credit slips before the customer returns the merchandise? You consider typographical problems (B6). That yields nothing. Then you try reading the sentence with various pauses (B4). This suggests the reading s’assurer dans le but rather than octroyer dans le but. But surely it isn’t the inspector’s job to make sure the goods go back … make sure the tobacco has been returned (you think, using a paraphrase (C8)). … Has been returned. Ah-hah! The customer’s possession of the credit slips constitutes the evidence that the tobacco has been returned—something a tax inspector needs to know (you realize from the rest of the text).

The writer made a poor syntactic choice: the deverbal noun retour does not indicate tense. Interpretation: in order to be sure that the goods have been returned, check that the customer has the credit slips from the manufacturer. (The trainer of inspectors who requested the translation confirmed this interpretation.)

Exercise 2

This is a question asked of candidates for a job that involves interviewing, along with the possible responses.

  • Pouvez-vous nous nommer cinq techniques d’entrevues?
    • reformulation, reflet, synthèse, résumé, établissement de liens, silence, demande d’explications, demande de prévisions, questions ouvertes, questions fermées.

Where does the problem lie? What questions should we ask to arrive at the right interpretation?

Answer. One of the responses—demande de prévisions—doesn’t make much sense. Is asking for a forecast an interviewing technique? (Perhaps there’s something you don’t know about interviewing, but we’re leaving that possibility out in this exercise). Perhaps there’s some other relevant sense of prévision (A3): expectation? estimate? Not yet willing to abandon the actual word the author used, you wonder whether the structure of the sentence (the list) might suggest something (D11). You notice that questions fermées is related to questions ouvertes, so perhaps demande de prévisions might be related to demande d’explications. A search in the WordPerfect thesaurus (C8) under explication yields précision as a synonym and you realize that the original text contained a typographical error.

Exercise 3

This is a passage from a report on an environmental impact study of Lake Saint-Louis, a portion of the St. Lawrence River. Can you find one problem in each sentence?

  • Sans pour autant restreindre l’importance et la nécessité de cette étude générale des différents écosystèmes du fleuve en fonction de la gestion des niveaux d’eau, l’étude d’une section d’importance de ce gigantesque écosystème permettra d’obtenir une connaissance valable d’une de ses constituantes. Le lac Saint-Louis se présente comme un des milieux hydrographiques et naturels très significatif du fleuve Saint-Laurent. Ceci, compte tenu des connaissances et données actuellement recueillies dans le cadre d’un projet de gestion des eaux de ce lac (projet Archipel).

Answer. The main clause of the first sentence is a tautology: the predicate is redundant (C9), adding no new information beyond what is in the subject. We learn that studying an important section of the St. Lawrence ecosystem (Lake Saint-Louis) will yield knowledge of a constituent (i.e. a section) of this ecosystem. The only information here is that studying something will yield knowledge of it, but this is almost a truism and surely not the author’s point. As to the knowledge being useful (valable), this adds nothing either since if there were no prospect of gaining useful knowledge, then there would be no point in mentioning the study. Once the tautology is noticed, the problem becomes one of target-language composition rather than interpretation.

The second sentence contains a minor anacoluthon, perhaps arising from editing with a word processor: it mixes un milieu très significatif with un des milieux les plus significatifs, but the meaning is reasonably clear: the difference between very significant and one of the most significant is not very … significant!

Finally, the function (D12) of the third sentence is not immediately obvious because the connector ceci is unclear. However, a second reading shows that the sentence is giving the justification for what was said in the preceding sentence: the results of the Archipel project suggest that Lake Saint-Louis is one of the most hydrographically significant portions of the St. Lawrence.

Exercise 4

What are the problems in this long sentence from a text about government-sponsored programs to ensure technological innovation in Canada and France?

  • Au Canada, effort d’association étroite de la recherche publique et industrielle avec des structures telles que l’IREM de Montréal, ou par la création d’une zone réservée dans laquelle sont hébergés les chercheurs des entreprises pour réaliser contractuellement avec l’IREM leur propre recherche et par la disponibilité sur le site, d’installations pilotes de différentes tailles permettant de simuler une production industrielle, créant ainsi les conditions d’une forte synergie entre recherche publique et recherche industrielle. En France …

There are three problems: failure to use commas to reflect sentence structure (B4), a serious typographical error (B6) and an anacoluthon. Corrected, the sentence would read:

… l’IREM de Montréal , par la création … leur propre recherche, et par la disponibilité, sur le site, d’installations pilotes … une production industrielle, on a créé les conditions d’une synergie  …

Au Canada:

  1. effort… l’IREM de Montréal


  2. par la création… et par la disponibilité. . . . . . . . . . .
[création d’une synergie]

Further Reading

Oddly enough, virtually no research appears to have been done—in English or French at any rate—on the subject of poorly written source texts (PWSTs) as they affect translators. The few short articles I did manage to find focus on the composition of the translation rather than the interpretation of the original.

Analysis of bad writing in the academic literature seems to be restricted to the diagnosis and remediation of the problems experienced by children learning to write, or college students learning to write better.

There are of course endless shelves of therapeutic books advising adults on how to improve their writing, or advising organizations on how to rewrite their documents in plain language. Both types of work give lists of problems, but with a view to avoiding or correcting bad writing rather than interpreting it.

It would be nice if the writers of the texts we translate had read some of these books, but they’re of little use to us as translators. Thus in Complete Plain Words, which I mentioned in the body of the article, the authors show how to rewrite numerous examples of bad writing by government officials, but they don’t explain how they arrived at their interpretations: they say this seems to mean that … or they simply assert (presumably after reading the rest of the document) that this means …

The present article is based on my own experience and intuition, but that is no substitute for systematic research on PWSTs, with a focus on how experienced translators deal with them. Most translation research concerns not the process but the final result of translation (perhaps in order to develop quality criteria). Recent studies dealing with the process tend to focus on the composing and revising of the translation (which are relatively ’visible’ activities) rather than on the work of research and interpretation.

As I was finishing this article, I did come across one book, from outside the field of translation, which sets out definite principles for interpreting problematic textual wordings:

  • Elmer Driedger, Construction of Statutes, Butterworths, Toronto, 2nd edition 1984.

Driedger, a former Deputy Minister of Justice Canada, draws on hundreds of court decisions to illustrate the principles judges use to determine what the law is. He offers the following approach to interpretation (p. 105) [my rephrasing and emphases—BM]:

  1. The Act as a whole is to be read in its entire context so as to ascertain the intention of Parliament (the law as expressly or impliedly enacted by the words), the object of the Act (the ends sought to be achieved), and the scheme of the Act (the relation between its individual provisions).
  2. The words of the individual provisions are then to be read in their grammatical and ordinary sense in the light of the intention, object and scheme. If they are clear and unambiguous and in harmony with the intention, object and scheme and with the general body of the law, that is the end.
  3. But if the words seem obscure or ambiguous, then a meaning that best accords with the intention, object and scheme is to be used, provided the words are reasonably capable of bearing that meaning.
  4. If on the other hand the words are clear and unambiguous when read in their grammatical and ordinary sense but the result is disharmony within the statute or with other laws, then an unordinary meaning that will produce harmony is to be given the words, provided they are reasonably capable of bearing that meaning.
  5. If obscurity, ambiguity or disharmony cannot be resolved by reference to the intention, object and scheme, then a meaning that appears to be the most reasonable may be selected.

As can be seen, these principles are very similar to those discussed in connection with Question 3 of Checklist Two. For specific rules of interpretation, see chapters 5 and 7 of Driedger.

There is of course a vast body of writing on the interpretation of literary—and especially sacred—texts. Special difficulties of interpretation arise with these texts because the reader’s knowledge of the real world cannot be invoked to the same extent as it can with legal, technical or administrative writings.

In English literary studies there is an exercise known as close reading, and in French literary studies an exercise known as explication de texte. These perhaps have potential for assisting with the PWST problem, though the manuals I was able to find did not show much promise in this regard.

One problem is the assumption that literary texts are well written; they are treated as what I called in Part One intrinsically difficult texts. A more general problem is that even pedagogical works on literary interpretation do not seem to set out principles for assigning a meaning to a difficult passage; instead they just give ’readings’ for specific texts. I managed to find only one exception:

  • Alan Durant and Nibel Fabb, Literary Studies in Action, Routledge, 1990. See especially chapter 7.

With sacred texts, the situation is somewhat better. While the wording in the original language is generally taken to be … well, sacrosanct, general principles of interpretation are given for dealing with obscure passages or passages for which there are variant manuscript versions.

Manuals for the interpretation—or exegesis as it is called—of sacred texts are of particular interest because some of them, in the Christian tradition at any rate, are written as aids to translators. One small drawback: the languages to which the techniques are applied will be Ancient Hebrew, New Testament Greek, Classical Arabic, Sanskrit and so forth. So if you are not familiar with these ancient tongues, the application of the interpretive principles will not be clear unless the writer provides interlinear glosses. A good place to start would be:

  • John Beekman and John Callow, Translating the Word of God, Zondervan, 1974. See especially chapters 17 to 19.

If you want to search for more materials in law or religion, keywords to bear in mind are construction (for law) and exegesis (for religion). Terms such as interpretation, hermeneutics and criticism are also used, but these are rather broad; most works keyed with these terms will not deal specifically or at any length with the linguistic problems that have been the focus of this article. Most interpretive writing focusses on the broader socio-political, historical-cultural or psychological-spiritual significance of the texts considered. What you should look for are chapters devoted to linguistic or grammatical or textual criticism.

Library of Congress classifications for browsing purposes:

  • Biblical exegesis and principles of interpretation: BS 476
  • Bible translation theory and principles: BS 449
  • construction of statutes: KE 482.S84


  • Back to the note1 This article is based on a workshop prepared for the Training Division of the Translation Bureau in October 1991 and on a paper read at the 2nd annual congress of the Canadian Association for Translation Studies, held in Quebec City in May 1989. All examples are authentic, though two or three have been modified to increase their illustrative value. Inevitably, passages which I found to be so poorly written as to be obscure in meaning will not be found to be such by all readers.
  • Back to the note2 Here are just two of the issues: Should we improve a text when the fact that it is poorly written could convey important information to readers about the writer? Should we eliminate bureaucratic jargon if the readers expect to see it and might wonder whether its absence indicates that the writer is not one of them?
  • Back to the note3 A hypertext is a collection of writings which can be read by following a number of alternative paths. The user browses through a passage, comes upon a point of interest, and then presses a key or clicks with a mouse in order to follow that point along its links to other portions of the collection of writings. The links are created using the programming language that comes with the software (e.g. HyperCard for Macintosh computers). Hypertext functions are also available in WordPerfect for Windows 6.0.
  • Back to the note4 Thanks to my colleague Gérald Jalbert for pointing out this trick and providing the example. His Translation Bureau workshop Traduction de textes mal rédigés was more focussed on the problems of drafting the translation than my parallel English-language workshop, which dealt primarily with source-text interpretation, but his notes helped me when it came to checking this article for completeness. For further discussion of how to use related words when interpreting a passage, see "The Role of Sense Relations in Translating Vague Business and Economic Texts" by Heidrun Gerzymisch-Arbogast, in Translation and Lexicography, John Benjamins, Amsterdam, 1989, pp. 187-195.