Translating and Copy-Editing
in a Commercial Environment

Commercial studies are usually narrowly focussed, and are necessarily based on small or unrepresentative sample populations. Nevertheless, they can suggest issues worthy of deeper research. The study reported here revealed serious deficiencies, in theoretical knowledge and practical tools, of a group of people who work in commercial translation. The central issue is the extent to which "language workers" — those for whom language is a significant element in working life — need training in formal grammar: the extent to which natural grammar, subconsciously acquired, is or is not sufficient without theoretical study. This issue bears on language education and on language policy. The paper offers some suggestions for in-depth formal research to reveal and measure the scope and scale of the problem.

Background

The origins of this paper lie chiefly in two unrelated episodes at Style Council 92. First, in a private conversation, a colleague expressed some concern about two papers which he felt were unsatisfying because they were based on rather small samples. Second, in a very public exchange, at least two other colleagues expressed serious doubt about the value of teaching "grammar".

Both views have some bearing on a commercial project that I was involved in at the time. The client was an Australian subsidiary of a Japanese corporation involved in high-technology manufacturing. Naturally enough, I am not at liberty to reveal the client’s identity, but that is not germane to the project and its findings.

I will discuss the project in some detail, but my purpose in doing so is merely to support my two points: first, that small surveys can function as "pilot studies", helping to decide whether an issue warrants a full-scale research project; second, that some knowledge of formal grammar is essential for anyone with a need to develop written material that is clear, concise, coherent, and completely comprehensible.

The Project

The client’s role — for present purposes — is to perform the final steps in the translation of technical materials from Japanese to English. Figure 1 illustrates the process.

Figure 1: The Process

This is a more complex process than one might expect: first because there aren’t enough really competent translators to cope with current volumes of international communication; second because the technical content of the material goes beyond the normal scope of second-language skills. Bell’s innovative study of translation theory (Bell, 1991) shows that a competent ("professional") translator needs full communicative competence in both the source and target languages, and also needs "real-world" knowledge that, when working with technical texts, is necessarily quite specialised.

Bell raises the question of the unit of translation, and advances the view that the unit is most often the clause (Bell, 1991: 29). As it happens, much of the text I saw appeared to have been translated very literally: the unit of translation appeared to have been the word. Given the frequency of idioms and other collocations in any language, it is not surprising that the translated texts lacked fluency. Since the translators were not native speakers of English — and may not have been native speakers of Japanese, fluency suffered in many respects. The problem was exacerbated by the fact that the Japanese parent company provided "word lists". In principle, these were lists of special terms and phrases with pre-determined translations. In practice, they often included fairly common expressions with wildly inappropriate but mandatory English renderings.

On the score of specialised knowledge, it is hardly surprising that many translators appear to have little understanding of technical terminology and concepts. For example, some of the translated texts showed confusion between cause and effect. This is a common problem, as the Japanese language encourages statements like "When A is true, B is true" — emphasising the co-occurrence without specifying causality. Without some understanding of the subject matter, the translators have no alternative but to guess, with only a 50% chance of being right.

The Japanese parent company developed this rather complex process in an attempt to overcome those problems of linguistic incompetence and technical ignorance in translators. Their idea was to use "native speakers" to tackle the linguistic problems, and to involve technically-competent people to weed out the technical problems.

My client employed three copy-editors whose task, in conjunction with technical people, was to perform the last three steps: to beat the often barely-comprehensible translated text into publishable shape. There was clear evidence, from a variety of sources, to suggest that they weren’t doing it very well.

My brief was to study what they do and how they do it; to identify problems in the translated texts and finished results; to identify correlations between problems (what original problems are fixed, what original problems are fixed inadequately or not fixed at all, what new problems are introduced); and to recommend whatever procedural changes or training initiatives seemed appropriate. A major outcome was that I developed and presented a course on Principles of Technical Copy-Editing. Course delivery was followed by further reviews to assess results and to see what problems the editors might be having in putting their new knowledge to use.

The following tables show my analysis of the problems at the three crucial points: as the editors received the draft, and at the end of their two edit passes. In each case, I simply took a passage of translated text containing fifty problems, then tracked that passage through the following stages to see what happened to the known problems and what new problems might be introduced.

A significant question was how to classify the problems. I tried several approaches, including such a detailed breakdown that each problem was close to being one of a kind. Finally, I adopted an analysis that considered how a problem could be detected and corrected as well as what kind of problem it was. The classifications are shown in the tables: the odd ".5" in some of the "corrected" values reflects a problem that was only partially resolved.

EDITOR A

Note: "-0.5" indicates a problem that was corrected in the first edit then partially reintroduced in the second edit

EDITOR B

EDITOR C

Overall results

The individual results showed some variation, but overall there was a fair degree of consistency. (Not that there was any point in computing means and variances with such a small sample!)

The problem categories need some clarification. In the following examples used to illustrate the categories, I have made minimal changes to ensure that the client organisation and its products cannot be identified.

"Lexical" problems were inappropriate or erroneous word choices. A typical example was numeral instead of number, as in a three-digit numeral.

"Usage" problems were expressions that, while formally correct, were simply "not English". For example, I found reference to a status indicator, where the status was said to be defined as effective; we would normally say something like in effect (or even on).

"Nominalisation" problems included both straightforward use of nouns instead of verbs, and the use of lengthy nominal strings. Both are too common in technical texts in general; in my client’s case, they were rife. An example of the first was This chapter describes how to ... perform preliminary checking for maintenance. A splendid example of the second was RF detector sensitivity threshold limit, which is admirably concise but rather difficult to interpret on first (or any subsequent) reading.

"Cohesion" problems were weaknesses in the structural integrity of the text. There were cases where it was difficult to work out what a pronoun referred to, and there were inconsistencies between subject and object selections in related sentences. An interesting structural problem appeared in an introduction: Chapter 3, "Settings used by AFE", describes the settings used to control AFE. The chapter title indicates that the settings are used by AFE; the rest of the sentence indicates that the settings are used by the user of AFE. Both are correct in their way, but used is being given two different senses. In consequence, the sentence seems to fall apart in the middle — it lacks structural integrity, or cohesion.

"Economy"’ problems were usually various kinds of verbosity, of which inappropriate use of the passive voice was the most common. Sometimes, though, the opposite situation prevailed: some text was too economical, causing the reader extra effort to fill in the gaps. An example of excessive economy was This section explains device control that uses AFE variable settings.

"Precision" problems included ambiguity of various types; incorrect choice of article; and use of the agentless passive. The Japanese language does not have elements corresponding to the English articles; context seems to do an equivalent job — but translators often manage to get it wrong, as in UMP is modification materials supplied to correct product faults. (UMP = Unit Maintenance Package.)

There were, of course, some overlaps between these categories. As pointed out above, inappropriate use of passive voice can cause problems in both economy and precision. There were in fact many cases where a single "error" could be classified in more than one way. In all such cases, the error was classified according to its most severe effect, or, where that was not possible, according to how it should be corrected. (Thus, for example, use of a transitive verb without an object might be seen as a syntax error; I preferred to class it as a usage error, because the solution is found by asking "How would we normally say this?")

Even in the worst case, where a passage appeared to be pure gibberish, it was essential to identity contributing problems. Clearly, it would have added nothing to our understanding of the process to label a passage as "gibberish" and to treat that as a problem category: there would always be more specific errors which could be tackled separately.

It may have struck the attentive reader that there is a surprising omission from these categories: there were no problems classified as "syntax". The translators produced text that was syntactically sound but otherwise quite dreadful; the editors rewrote much of this text without creating syntactical errors, but they did create other problems. This lends support to the view that we don’t need to learn formal grammar in order to speak or write grammatically: these three copy-editors had — as I shall describe — virtually no knowledge of formal syntax, yet their work was free of syntactical error.

To summarise, it appeared that there were two real sources of problems. First, the translators were working at a lexical rather than a semantic level, producing unsatisfactory drafts. Second, the copy-editors lacked the skills to identify specific problems and develop appropriate solutions. We could not do much about the translators; we could work on editorial skills.

The Training Program

As I have already mentioned, in response to those findings I put together a short course. In outline, the topics covered were Objectives of Editing (Ease of understanding, Accuracy of understanding, Factors involved in understanding); Problem categories and concepts; Golden rules and old wives’ tales; When passive is right; Schoolroom rules that are best forgotten. Case studies were based on a variety of sample passages.

When I presented the course, I was pained and surprised to discover how little the participants knew about the mechanics of language. Here were three people who had chosen to earn their living not only as language workers — in the sense that journalists, for example, are language workers — but as authorities: people who asserted the right and claimed the expertise to improve the written language of others. Their lack of knowledge appeared to be a major cause of their poor results; but, more to the present point, it prevented them from even exploring those problems. I need hardly add that it was extraordinarily difficult to explain (for example) how to eliminate a misplaced modifier, when they could not recognise it in the first place and could not see why it was a problem in the second.

Conclusions

I draw two conclusions from this exercise. First, a knowledge of formal grammar is of great importance — is perhaps essential — for people whose business is language. Second, the inability to analyse texts in a rigorous way is probably widespread in the community.

(The last point has been borne out by my experience as a teacher in the Student Writing Skills program at Macquarie University. Of course, I do not claim that this evidence is anything better than anecdotal, but it is at least consistent.)

I have already acknowledged that we do not need training in formal grammar in order to speak or write grammatically: formal grammar is not a tool for language use. Rather, it is a tool for analysis. When natural grammar tells us that something is amiss, we generally need formal grammar to work out just what is amiss and how to fix it. I believe that the relationship between natural and formal grammar is as shown (albeit simplistically) in Figure 2. L1 speakers are able to use their language both grammatically and fluently, but, unless they have formalised their grammatical knowledge, they are unable to analyse texts. L2 speakers may be able to point out why a text is ungrammatical, but, unless they have internalised their grammatical knowledge, they lack genuine fluency. In short, natural grammar provides the ability to create texts which display fluency; formal grammar provides the ability to analyse texts which lack clarity or are otherwise deficient.

Figure 2: Natural and formal grammars

Needless (I hope) to say, I do not advocate that ancient shibboleths be drilled into our schoolchildren. Rather, the recent decision by the NSW Board of Studies to introduce functional grammar into the school curriculum is to be applauded: this is a formal approach to how we use language to communicate, rather than to how constituents are glued together — but it does provide the tools for rigorous analysis and for the repair of textual defects.

I have suggested that the inability to apply rigorous analysis to texts is probably widespread. I can be no more certain or categorical than that, simply because the conclusion is based on a very small sample. Nevertheless, the findings are consistent (though not uniform) within that population. In short, the study I have described was indicative rather than definitive; a definitive study would have required more resources than I or my client could bring to bear — and would have been irrelevant to my client’s needs. But large-scale studies do not always produce interesting or useful results. This small-scale study has given us some strong evidence to suggest that, in this case, the results of a large-scale study would be both interesting and useful.

The debate about the teaching of grammar has been resolved, at least in NSW, and at least for the time being. I believe that we now have a golden opportunity for some significant research into the value of such teaching, by running multiple parallel studies along the same lines as the one I have described. Conducted in different states, amongst various occupational groups, and repeated at appropriate intervals, these studies could help to determine conclusively whether the teaching of functional grammar does enable people to write more effectively — with greater clarity and comprehensibility.

 

REFERENCE

Bell, R.T. (1991): Translation and Translating. London: Longmans

If you have questions or comments about this web site, please send them to us.
Copyright © Brandle Pty Limited, 1997-2006
Last modified: March 24, 2002