Artikel van Pierke Bosschieter verschenen in Information Wissenschaft & Praxis 8/2007
Translate the index or index the translation?
In this article the author describes the procedure of providing page numbers for a verbatim translated index. She explains the difficulties she encountered while compiling such an index. With this experience in mind she strongly argues against this method, in favour of indexing from scratch.
Publishing of translated non-fiction books
Many newly-published non-fiction books in Holland and Germany are translated from English. Statistics are hard to come by. In 2006, of the 81,177 newly published books in Germany, 3,781 were translations from the English, but according to the brancheninfo these were mostly novels. Taking the Royal Dutch Library, NetUit, Website as my basis, I tried to establish some figures for the Dutch publishing industry. Looking at the A-list and only at non-fiction, I came up with the rough estimate that about 7% of the non-fiction books published in the previous couple of months were books translated from English. Even more telling was that some of these books had indexes in the original publication but not in the Dutch version. As I only checked a few on Amazon.com, with the ‘Look Inside this Book’ feature this is not a well-researched statement and should be read with caution. I could not find any total figure for books published in Holland later than the 2002 figure of 19,061.
In the absence of firm figures, for the purposes of this article, I will assume that the translation of English non-fiction books is a normal occurrence in both countries and is done on a regular basis.
The Indexing Profession in Germany and The Netherlands
In her article Continental European Indexing, Caroline Diepenveen makes some general remarks about indexing practice in Germany and The Netherlands. Her researches suggest that about 52.5% of non-fiction books in Holland have an index, approximately the same percentage as Martin Tulic quotes for the US in the first quarter of 2006. Quantitatively equal but qualitatively not. Over the last 50 years or so, Anglo-Saxon countries have developed indexing into a science (if not an art form), while on the European mainland indexing has had little attention. In Great Britain, as in the USA, there are many ways open to students to follow specialized training courses at private institutions or universities. In addition, indexers in the English-speaking world can turn to their national indexing societies for help with on-going professional development and for general professional support.
Germany and The Netherlands now both have their own indexing networks, the Deutsches Netzwerk der Indexer (DNI) and the Nederlands Indexers Netwerk (NIN), which are signatories, as fully-fledged affiliates, of the International agreement of indexing societies. Both networks hope to draw attention to the indexing profession and to raise the standard of indexing in the two countries.
Why opt for a translation
In English-language books, an index usually is a subject or analytical index, while indexes in German and Dutch publications are often poor quality indexes or just names indexes. In addition, these indexes usually have long strings of page numbers attached to each entry, which is not exactly user-friendly. For the reader of a non-fiction book a general subject index is of course much more helpful than a subject index with long strings of locators or an index containing only names of persons.
One of the first things publishers deciding to publish a translation of an English or American non-fiction book come up against is the realization that the index in the original book is much more helpful than the average Dutch or German index. Not knowing much about the indexing process and unaware of the professionalism of the indexers in the German and Dutch Networks, some publishers choose the option of verbatim translation of the index. This choice is often prompted out off reverence for the book (and its index) and/or because of financial constraints. They assume that this method will produce an index of the same quality as the English index, as cost-effectively as possible.
What publishers should be aware of is the intricate process of writing an index. A professional indexer starts reading on page one, works his or her way through the complete text and while doing so analyses the text and picks out the concepts, subjects and names that should be included in the index. This sounds simple and straightforward, but it is actually a rather complex intellectual and creative process. After identifying the indexable matter, the indexer needs to find terminology for the index that is appropriate to the prospective readers of the book. It’s not a matter of merely underlining of important names and keywords in a text and putting these in alphabetical order, for that could be done by a computer. Because of syntactic and semantic ambiguity, the understanding of a text is far from straightforward. Think of the use of synonyms and homographs, of the fact that certain concepts are not implicitly mentioned and of the fact that sometimes the less important words hold the key to understanding a text. That is the indexing challenge.
In addition indexers have to adhere to rules, formulated in indexing standards and intended to help the user, about the construction, lay-out, typography and other technical details of an index. The Dutch standard NEN 3547 was published in 1981 and withdrawn in 2001. Since then the official standard in Holland is the international ISO 999 which was published in 1996 and is widely used as an authoritative guide worldwide. Germany, on the other hand, has the DIN 31630, which has been published in 1988. According to Jochen Fassbender the standard covers the important aspects of indexing, although some of it is now out of date (for instance the recommended use of the abbreviations f. for following page and ff. for following pages). Moreover, it does not cover a number of important aspects, such as the limitation of the number of locators or page numbers to be attached to a heading and it does not say anything about passing mentions, a subject I will return to later.
A case study:
In August 2007 I was asked by the Dutch publisher Nieuw Amsterdam to provide page numbers to the verbatim translation of the index to ‘The Years of Extermination 1939–1945, Nazi-Germany and the Jews’ by Saul Friedländer, which was published to much acclaim in the US in 2006 by HarperCollins. It is the second and last part of Friedländer’s comprehensive study of the Holocaust and the English version has about 665 indexable pages. This was the first time I was ever offered such a job and I thought it would be a nice challenge, especially since I had to deliver the index in a week.
A German version of the book had already been published by C. H. Beck Verlag, at the beginning of 2007. Both the American and German versions have an index.
Nieuw Amsterdam had the index translated in parallel with the translation of the text, by a different translator. This second translator had a solid knowledge of the subject and kept in touch with the translator of the text. After completing the translation of the English index, he compared it with the German index. Entries that appeared only in the German index were duly translated and added to the ‘master list’. The idea behind this was that they would now have a comprehensive ‘master list’, equal the level of detail of the book. It would also, thought Nieuw Amsterdam, be a time- and money-saver, for on completion of the Dutch translation, this ‘master list’ needed only page numbers to turn it into a first-rate index.
I was presented with a ‘master list’ of 2763 lines, each line representing one main entry or subentry. The list was not in alphabetical order and only the entries copied from the German version had page numbers attached (of course referring to the German text). I was also provided with PDF files and hard copies of the book in all three languages. As I said earlier, an index cannot be generated automatically by computer, but the computer can greatly assist the human indexer. There are three major software packages available at the moment: CINDEX, MACREX and SKY Index. As I use SKY, my remarks about software are mainly references to SKY. SKY’s main window is divided horizontally in two; the upper part showing the index in progress, the lower part showing a database grid where you can enter and modify your entries. The software has powerful tools to help with building an index, including fast data entering, tools to help with sorting order and repagination, lay-out and typography, and tools for checking inconsistencies.
I started by loading the ‘master list’ into SKY, so that I could use all the alphabetization and lay-out tools. I then made a transposing chart, listing the page numbers of the English edition with the corresponding page numbers of the Dutch edition. Of course this was a rough chart, because I couldn’t include all 661 page numbers, for lack of time and because the list would have become too unwieldy. The list fitted an A4 page and hung below my main screen. Another page, also fastened to the main screen, had a list of corresponding page numbers of the chapters and paragraphs of the Dutch edition. My big main screen had an open SKY window and Adobe reader with the Dutch text. On my second screen I had the second Adobe reader with the English PDF file opened. In addition I had printed versions of the ‘master list’, the English index and the German index on my desk in front of me. Then I took the ‘master list’ and started with the first heading.
What was of course obvious from the start was that it was easy enough, using the search facility in Adobe Reader, to find names of persons, geographical entities and corporations. But what became clear when working my way down the ‘master list’ was that the names inserted from the German index were very often passing mentions and therefore omitted by the American indexer. It is one of the discrepancies between the DIN 31630 rules and the ISO 999 rules. According to the latter, an indexer should discriminate between information on a subject and the passing mention of a subject and exclude passing mentions that offer nothing significant to the potential user.
Another significant difference in the indexing practice of English and German indexers, relates to the use of long strings of undifferentiated page numbers. Although ISO 999 states that users should be able to retrieve information quickly, the DIN 31630 doesn’t touch on this subject at all. Leading American and English literature on indexing clearly explains the usual upper limit of 5 or 6 locators per entry. In fact both indexes had long strings of locators. In the German index it was a very common occurrence, in the American index an exception.
As the translated index can only be as good as the original version, the publisher should have ascertained the quality of the indexes, before he decided to have them translated. The German one had a very simple structure, made no use of see and see also references, used very long strings of locators, made use of ‘f.’ and ‘ff.’(which is not recommended by ISO 999 because it gives the user incomplete information), contained a lot of passing mentions and had too few subheadings. On the other hand, the American one had a lot of merits, like an intricate structure with see and see also references, an appropriate number of subheadings and no passing mentions. In this case, the American index seemed on the face of it, to be a good one and should not have been augmented by mostly unnecessary headings from the German index.
The structure of an index is very intricate and develops slowly as the indexer makes his or her way through the text. Usually the structure is created simultaneously with the selection of main and subheadings, so that interrelated terms can be created on the fly. This can mean that one page reference may be attached to many headings and sub-headings. For example:
in Belgium, 121-122
in Holland, 121-123
Jewish council and, 121
In Nazi-soldiers’ letters, 121
Anti-Jewish measures in, 121-122
Bieberstein, Wilhem, 121
Comité of Coordination, 121
Comité de Défense des Juives, 121
Danneker, Theodore, 121
anti-Jewish measures, 121-124
in France, 121
in Holland, 121
letters, German soldiers’
anti-Semitism of, 121
The above example shows the interrelationships between headings created when the indexer read page 121. It shows that when Belgium is mentioned under anti-Jewish measures, the main heading Belgium should have a subheading anti-Jewish measures and the page range for both should be identical. The same applies to anti-Semitism in Nazi soldiers’ letters. While working through the ‘master list’ I discovered that the American index wasn’t as well structured as I first thought, for a lot of the above mentioned interrelationships were missing or the page ranges for some of the interrelated terms were not identical in instances were they should have been.
My example also makes it abundantly clear, why compiling an index by simple translation, is the most ineffective method imaginable. For an indexer working from scratch reads page 121 once and makes the 15 entries and moves on to page 122. Working with the ‘master-list’ I did encounter the page references to page 121 fifteen times and had to scan the page anew each time to see where a concept ended and began. Coming from any number of other pages I had to familiarize myself yet again with the content of page 121.
My finished index has 5,300 locators. I checked at least 1,000 locators that turned out to be passing mentions and didn’t end up in the finished index. The book had about 625 indexable pages. So in the end I looked at each page an average of 10 times. In addition, although I had the transposing list to work from, it was time-consuming to find the exact passage the page reference was referring to, especially so with underlying concepts and ideas like ‘aangeven van Joden’ (denouncing/betraying Jews). I had also to repeatedly read/scan the next pages to see where the passage ended.
Another aspect that slowed me down was the fact that the ‘master list’ contained words that simply did not appear in the text, due to the fact that the translator of the index was not the translator of the text.
In the end it took me 100 hours to complete the index. Because of time constraints – six working days – the final index still had a lot of main headings with long strings of page numbers attached. The structure was better, but not flawless and there wasn’t enough time to sift out all passing mentions. Indexing from scratch, taking no more and probably much less time, would have resulted in a higher quality index. And as there would have been no need for translation, this method would also have been rather cheaper.
An afterthought about copyright
Who is to take the credit or the blame for this translated index? The original indexer, the translator of the index, or me, the ‘minion’ who dealt with all the drudgery of converting the page references? That is to say, I don’t believe this latter job could have been done without a thorough knowledge of indexing.
Even re-indexing from scratch with the original index as a reference, can in practice result in an index close to a verbatim translation, in which case, the original indexer might also have copyright claims, but these will be extremely difficult to enforce.
And what if the end result is a very poor index (which I hope is not the case with the Friedländer book)? Aside from any questions about copyright, this might be a breach of moral rights, since the resulting index could lay the original indexer open to ridicule.
The translation of an existing index may well seem the easy option, but in practice, because there is no automaticity about the indexing process, translating an index will probably run into many problems. If the original index is a poor or mediocre one, the quality of the translated index won’t surpass it. Editing the translated index requires the skills of a professional indexer, and carrying out the necessary checks is likely to be both tedious and time-consuming. Bearing this in mind, it is far better to start from scratch.
Diepeveen, Caroline (2006) Continental European indexing: then and now. The Indexer 25 (2), 74–78
Deutsches Institut für Normung (1988) DIN 31630 (Teil 1) Registererstellung: Begriffe, Formale Gestaltung von gedruckten Registern.
Fassbender, Jochen (2006) German indexing: some observations on typographical practice. The Indexer 25 (2), 79–82
The Indexer 25 (3) Centrepiece 2, C16
International Standard, ISO 999 (1996) Information and documentation: Guidelines for the content, organization and presentation of indexes.