November 10 2006
Sir, -Dinah Birch was justified in praising Philip Waller's Writers, Readers, and Reputations: Literary life in Britain, 1870-1918 (September 1) and equally justified in the theme of her review: "This is an extraordinary mine of fact, detail, quotation, anecdote and reminiscence". Those who use the book, however, will feel the want of a miner's lamp. I looked my own name up in the index, to find only two references. Reading Waller's 1,046 pages of text was more reassuring. There are some fifty or more places where (with due acknowledgement on the page) work of mine was cited, drawn on, and -in some places -corrected. Other Victorian scholars with whom the text engages are similarly blanked out in the index. To take one, of very many: on page 668 Waller footnotes his text with the graceful acknowledgement: "Dr J. J. McAleer kindly supplied me with this reference". There are other points in the text where Dr McAleer's work is used (with on-page acknowledgement). But the name is missing from the index.
Chapter Five is titled, allusively, "The Great Tradition" and opens: "When the Cambridge literary scholar F. R. Leavis published The Great Tradition in 1948, he was adding the coping stone to a structure of Victorian and Edwardian construction". The discussion of the Leavises' austere economy of textual reference -which Waller's enterprise, with its vast extensiveness, contradicts -goes on for three pages. It is an important statement of critical principle. Writers, Readers, and Reputations is an anti-Fiction and the Reading Public.
But there is no entry for either Leavis in the index. The index (or lack of indexing) critically disables a superb book. "No one", says Dinah Birch, "could duplicate (Waller's) reading." But how can lesser readers find the needles in this vast haystack? Even where there are index references, they are mainly in that most useless form: page numbers (around a hundred, for example, on "Great War 1914-18": what use is that?).
Every year the Society of Indexers, of which I am currently president, criticizes the modern form of publishing contract imposed on scholars because it throws on the author the entire responsibility for the index. What this typically means is a few exhausted days with the page proofs; and indexes of the kind that limit the usefulness of Mr Waller's book. This book was a quarter of a century in the writing. It is immensely long and, at Pounds 85, painfully expensive. A haporth of tar (in the form of a worthy index) was, I believe, a false economy on the part of the Oxford University Press.
JOHN SUTHERLAND
Department of English, University College, Gower Street, London WC1.
November 17 2006
Sir, -John Sutherland's plea for more effective indexing of scholarly books struck a chord, not least because I must confess to precisely the sin he identifies, of preparing my own index in a rush at the last minute. In mitigation, since the alternative was to spend a sizeable proportion of the likely royalties from a specialist academic book on professional help, I succumbed (sadly) to avarice. Perhaps beleaguered academics might form an alliance with equally beleaguered indexers to persuade publishers to bear what (for them) would be a relatively small cost in return for a markedly better product? By the way, my book is entitled Wisdom Literature. Sadly ironic, perhaps.
ALASTAIR G. HUNTER
Department of Theology and Religious Studies, University of Glasgow, Glasgow.
Sir, -The solution to the problem of books being published without adequate indexes that John Sutherland identifies (Letters, November 10) appears elsewhere in the same number of the TLS. Mark Kidel describes the new Bob Dylan Encyclopedia being sold with a CD containing an electronic text of the whole work. This has long been the practice with books in the field of Computer Studies and it obviates manual indexing, since any word or phrase may be found by the software. Indexing is a mind-numbing task made necessary by the limitations of paper publishing, and the author who skimps on it can be forgiven since technology will soon make it pointless even for print-only works. The recent works of the author in question, Philip Waller, have been digitized by the Google Books project and the full text indexes made publicly available over the internet, and we may expect the same of his latest work unless the author or publisher tries to prevent it. Of course, it is regrettable that the mass digitization of books is being undertaken by a commercial organization rather than by public bodies. The fault for that lies with libraries, authors, and most especially publishers who continue to erect barriers to digitization.
GABRIEL EGAN
Department of English and Drama, Loughborough University, Loughborough.
November 24 2006
Sir, - Gabriel Egan says that indexing is a mind-numbing task. It can be a pleasure if the author uses the elegant software MACREX.
(www.macrex.com). Not only does it facilitate indexing, but it enables the bulk of the work to be done from the typescript, avoiding the problem of the few exhausted days with the page proofs which John Sutherland wrote about (Letters, November 10).
MICHAEL NELSON 21 Lansdowne Road, London W11.
Sir, -Gabriel Egan's suggestion that digitization and full-text searches remove the need for book indexes is reminiscent of the enthusiasm for automated natural language translation when computers first arrived -that resulted in such famous examples as "the spirit is willing but the flesh is weak" becoming "the wine's OK but the meat's off". Forty years on and still no one would trust a machine-translated document without human post-editing. Similar problems face automatic indexing.
Computer production of a concordance, showing the location in the text of each word or phrase, is enticingly easy and gives the paper equivalent of a full-text search facility, as is common on CD-ROMs or internet search engines, yet provides only a starting point for finding information.
Language is full of homographs and synonyms and not all the references will be relevant. For example, a computer-produced index of the 9/11 report showed many references for George Bush, but did not distinguish between GB senior and George W. In an environmental text, the phrase "lead users" will refer, not only to users of the metallic element, but also to early adopters of technology. You may find time-wasting references, such as looking up "teaching children" and finding only the statement telling you ". . . the above is not relevant when teaching children". Equally concerning, you may be missing information -looking up the Dardanelles and missing references to the Hellespont or Canakkale Bogazi, or seeking information about population and missing discussions about the number of people living in the area.
A human-produced index has had someone check each and every part of the text to find everything relevant to the search term -the indexer has done the work, whereas a concordance pushes the responsibility for finding the information back onto the enquirer. There is no doubt that the enquirer is better served by a good, human-produced index than a full-text search. Whether producing an index is mind- numbing probably depends on the book concerned, but if an author finds going through their book, yet another time, in meticulous detail, stupefying, surely they should call in a professional rather than disadvantage their readers.
JAMES LAMB Broadview, John de Bois Hill, Ardleigh, Essex.
Sir, -The letter from Gabriel Egan (Letters, November 17) displays an extraordinarily limited view of the nature and functions of an index, one of whose essential characteristics is, and always will be, structure. The entry for David Garrick in the index to the Oxford edition of Boswell's Life of Johnson (an example chosen almost at random) has more than a hundred subheadings, directing the user to such topics as Garrick's conversational powers, his performance in Hamlet, physical appearance, etc. All of these could be found via an electronic text of the whole work, but only by inputting "Garrick" as a search term and checking the relevance of each and every occurrence of the name, a laborious and time-consuming procedure which the provision of a structured index renders quite unnecessary.
C. P. RAVILIOUS 1 Goodwood Close, Eastbourne.
Sir, -John Sutherland (Letters, November 10) rightly deplores the practice of making scholarly authors entirely responsible for the indexes to their books.
Indexing is a skilled process which in most cases, and particularly in the case of complex texts, should be undertaken by an experienced professional. A good professional indexer will, for example, avoid the beginner's classic fault the long strings of undifferentiated page references of which Professor Sutherland complains in the index to Philip Waller's Writers, Readers, and Reputations.
TLS reviewers often mention indexes, and their impact on the value of the book being reviewed, whether good ("excellent index", April 28, 2006), bad ("index ... wholly inadequate", February 24, 2006) or absent ("There is, alas, no index, just ten blank pages at the end, and this makes the book less valuable as a work of reference", May 19, 2006). Publishers who wish to avoid adverse comments might well consider altering their contracts to take the onus for preparing the index away from the author, who is nearly always unqualified for this task. Meanwhile, harassed authors seeking professional help would be well advised to consult the directory on the Society of Indexers' website (http:// www.indexers.org.uk).
CHRISTINE SHUTTLEWORTH
Flat 1, 25 St Stephen's Avenue, London W12.
December 1 2006
Sir, -Following the interesting exchanges on indexes and indexing, I was reminded of Emerson writing in the spring of 1847 that "Among the seven ages of human life, the period of indexes should not be forgotten".
WOLF LEPENIES
Wissenschaftskolleg zu Berlin, Wallotstrasse 19, 14193 Berlin.
Sir, -Having made indexes for three books of my own and one written by a friend, all on medieval history, I find it hard to see how a professional indexer could either have done the job in each case in the time likely to be allowed by publishers, or could have known enough about the contents to cover subjects as well as I did -which is not to claim that I did it as well as readers of the books might wish.
The author must know better than anyone else what are the important subjects to be indexed. Even telling a professional indexer what they are may not make all references to them easy to recognize on the page. As for computer programmes, it is difficult to see how well they can do before they are able to index ideas as well as words. This problem applies to indexing people as well as subjects.
References to people by their title need to be connected to the right king, count, bishop or whatever. Doing this for my friend's book, which was on a different period and aspect of medieval history from my own, took more time than professionals earning their living by it could possibly spend, even if they had access to the sort of specialized open-access library it needed. Then there are places, possibly in other countries, that should ideally be identified and located by country and province etc.
None of this means that there is not a place, and an honourable place, for good professional indexers. Different kinds of books pose different problems. Nor is it always possible for authors of scholarly works to do the jobs themselves: there will always be special cases. But the making of one's own index ought surely in principle to be part of a scholar's obligations to one's subject and one's possible readers.
SUSAN REYNOLDS
Institute of Historical Research, Senate House, Malet Street, London WC1.
Sir, - John Sutherland rightly presses for a good index (Letters, November 10), but shouldn't authors and publishers be crowding into your columns to protest against his idea that authors should not undertake "the entire responsibility" for it? He assumes that indexing occurs only in the "few exhausted days with the page proofs", yet this is to sacrifice half the index's value. Drafting it earlier from typescript (whether computerized or not) enables the author to eliminate errors, contradictions and anomalies from his book before they are frozen into semipermanence at the proof stage. The addition of the final page numbers can then occur during the "few exhausted days", when far fewer proof corrections will be needed.
And whether we use the MACREX computer indexing package or not, shouldn't we all be backing Michael Nelson (Letters, November 24) in dissenting from Gabriel Egan's view (Letters, November 17) that "indexing is a mind-numbing task"? On the contrary, it is a creative activity which enables the author to bring out a book's full richness by rearranging its contents in ways that differ from, but supplement, the arrangement chosen for the main body of the book. The skilful use of subject headings is especially important here. There is no such thing as a "professional" indexer apart from the author, who alone possesses the essential qualification of knowing what he wants to say, and who has every interest in clarifying his message by fully differentiating his page references.
BRIAN HARRISON
The Book House, Yarnells Hill, Oxford.
December 8 2006
Sir, -The index omissions that prompted John Sutherland's complaint (Letters, November 10) were names of people, which even the simplest machine can find. Those who wrote to the TLS (Letters, November 24) to insist that an index is semantically more complex than a mere concordance are right and should be delighted at what machines can now do.
The current state of the art in natural language processing enables software to replicate the systematic ordering of terms and ideas found in a manually produced index and to greatly outperform it for sophisticated enquiries of the kind "where does this author mention A or B without also mentioning X or Y?". To take a concrete example, the Nameless Shakespeare computer project (http://wordhoard.northwestern.edu) allows one to ask the machine to list just the adjectives used by the character Ophelia when speaking verse, and to identify which of them are not also used in, say, Desdemona's or Imogen's verse lines. Those persuaded by James Lamb's quotation (Letters, November 24) of laughably poor machine translation of idiomatic English that machines cannot make sense of language should note that the morphosyntactic analysis necessary to tell Ophelia's adjectives from other parts of speech was done by machine, not by hand. Thus the computer sets the mind free to pursue details that were formerly too difficult to discern. In the way they mistake the impact of technology, those who insist that readers will always prefer well-made manual indexes are like the medieval scribes who, confronted by the crude but rapid early printing press, comforted themselves that serious readers would always prefer painstakingly illuminated manuscripts.
GABRIEL EGAN. Department of English and Drama, Loughborough University, Loughborough.
Sir, -A good index will judge the significance of topics, then analyse, structure and distinguish them -or, rather, the good indexer will. Text searches are a poor surrogate, because they neither understand nor discriminate. Author-indexers know from experience how hard it is to produce an accurate, comprehensive and well-judged index, which is why professional indexers exist. They bring a fresh eye, long experience and a well-stocked brain.
As Alastair G. Hunter says (Letters, November 17), publishers should not expect academic authors to index their own books; potentially, authors may make the best indexers, but they cannot devote all their time to the task and all too often their inexperience shows.
I suspect Brian Harrison (Letters, December 1) knows that professional indexers do exist (I am one) and that, unlike the author, "who alone possesses the essential qualification of knowing what he wants to say", they index what the author actually said, which is more useful to the reader. Susan Reynolds (December 1) takes a delightfully dim view of indexers' abilities. Would they, she asks rhetorically, be able to identify "the right king" and "places, possibly in other countries"? "Yes" is the short answer; these are routine problems. My index to Ancient Rome differentiated eleven men called Appius Claudius -and the author was dead, so he couldn't help.
Medieval history is not a problem either, if you go to a suitable person: academic authors have commissioned me to index books on topics as far apart as the Burghal Hidage and the South English Legendary. There are professional indexers in every specialist area, as an online search for "indexers available" will show.
GERARD M-F HILL. Chapel House, Beckstonegate, Low Row, Brampton, Cumbria.
December 15 2006
Sir, - Gabriel Egan (Letters, December 8) confuses the issue: the Nameless Shakespeare project does an excellent job of analysing linguistic features, but would not have helped Gerard M-F Hill (ibid.) distinguish his Appius Claudiuses.
The book I am indexing at the moment often refers to scholars by their surname alone; for the forenames or initials I have to hunt on other pages, or earlier editions of the book, or library catalogues or works of reference. I do not see how natural language processing could help in this task of identifying and distinguishing people. Nor does it seem likely that a computer program could take a string of thirty undifferentiated page locators, consider the options for intelligently discriminating between them, and offer the analytical subheadings which will be most helpful for the reader. Indeed, while the computer may be good at finding every occurrence of a term, that is not always what is required: in most contexts, readers are not well served by an index which has them looking up every passing mention of a subject, while ignoring substantive discussions of a key person referred to only as "him" or "his son".
CHRISTOPHER PIPE. Watermark, Norfolk House, Cromer.
December 22 2006
Sir, - I am reluctant to prolong the correspondence about indexing (December 15), but feel I have to try to correct a misunderstanding about indexing software like MACREX, which has been evident in many of the letters. The indexer does not use the book in machine-readable form. The entries are put into the computer manually from the hard copy. Therefore the indexer is free to introduce ideas as entries even if they do not appear as words in the text of the book. The purpose of the software is to facilitate the process of indexing and automate many of the otherwise tedious repetitive tasks. When the proofs come back from the publisher, the indexer turns the index from alphabetical to page-number form with a simple instruction and changes its page numbers to match those in the proof.
MICHAEL NELSON
21 Lansdowne Road, London W11.
