It is with some trepidation that I leave the train at Oxford and stride off towards the offices of the Oxford English Dictionary in Great Clarendon Street. I imagine the scene that awaits me: dark, oak-panelled rooms occupied by crusty, bearded old pedants whose chief pleasure in life consists of heaping scorn on people like me for their sloppy use of the English language.

“People like you? People like you?” I hear them exclaiming already. “Presumably, you mean people such as you. If they were like you, they would have to bear a close resemblance to you, which would mercifully limit their number.”

Oh dear, am I on the cusp of embarrassing myself? “Hardly,” I hear them sigh in exasperation. “Did you sleep through Latin at school? Cusp comes from cuspis, meaning point or apex. On the cusp means being at the very peak of doing something - almost the opposite of being on the verge, if that’s what you’re trying to say.”

Well, can we at least agree that I am putting myself in the firing line? “Oh, for pity’s sake,” I hear them groan. “Do you mean in the line of fire? Surely you don’t need a dictionary to tell you that the people in the firing line are the ones doing the shooting?”

As I turn into Great Clarendon Street, my anxiety grows. The headquarters of the Oxford University Press, publisher of the OED, turns out to be a predictably imposing 19th-century edifice built in the style of an Oxford college around a quadrangle. It oozes history, erudition and authority.

So the first surprise on entering is to find that the OED offices, housed in an annex to the main building, look much like offices anywhere: modern, open-plan and liberally scattered with desktop computers. The atmosphere is relaxed, and most of the 60 or so editorial staff scattered around the place, far from being old and crusty, look disarmingly youthful.

The second surprise is to meet John Simpson, the dictionary’s 53-year-old chief editor. Unassuming and softly spoken, he smiles as I open my notebook and begin my clumsy scrawling. “Ah, I’m glad to see you’re left-handed,” he says. (He is too.) “There are more left-handers in lexicography than in the general population. I don’t know why. It’s the same with computing - something to do with being logical and analytical, perhaps.”

Simpson makes it clear right away that anyone looking for pompous didacticism has come to the wrong place. “We’re descriptive, not prescriptive,” he says. Unlike the official dictionary of the Academie francaise, Fowler’s Modern English Usage, or indeed the FT’s in-house style guide, the OED makes no attempt to instruct or advise on the correct use of the language. It simply records the way English words are actually used, and if a word is misused or misspelled long and often enough, the new meaning or spelling finds its way into the dictionary, however much the linguistic puritans may protest.

Shocking as some may find it, “phenomenons” and “phenomenas” are listed as plural forms of “phenomenon”, though the dictionary gently notes that “phenomena” is “normally preferred in modern use”. Excruciating misspellings such as “mischievious” for “mischievous” and “contractural” for “contractual” are also included, as is the misuse of “flaunt” for “flout” in phrases such as “flaunting the law”. None are condemned, merely noted as non-standard.

And what does the dictionary have to say about “hopefully”, one of the most reviled English words of modern times? Purists deplore the way a perfectly good adverb, as in “better to travel hopefully than to arrive”, is now used as ungrammatical shorthand for “it is to be hoped that”, as in “hopefully, it won’t rain tomorrow”. (The adverb, you will observe, is left dangling, without a verb or any other word to modify.) The OED, however, provides a full entry for the disputed sense of the word, pausing only to mention that it is “avoided by many writers”.

Asked how he feels about people who lament declining standards in the use of English, Simpson says: “I’m not particularly sympathetic to the old-fashioned Disgusted of Tunbridge Wells critics of English.” For him, they call to mind the so-called etymological fallacy, according to which words have a correct meaning fixed at a particular point in time and dictated by their etymological origins. Based on that principle, he points out, “forensic” could still only mean of or pertaining to the Forum, the Latin word from which the adjective is derived, instead of being a useful modern-day word for material produced in court for evidence.

Unlike French, Simpson says, which French-speakers see as a purer language, English has been in a perpetual state of change and evolution throughout its 1,500-year history, eagerly absorbing words from other languages and spawning new versions of itself in North America, Australia, the Caribbean and the rest of the English-speaking world. “We’re prepared to be more flexible than many other languages are,” he says. “To some extent, the OED’s descriptivism mirrors the English-speaker’s approach to the language.”

It is exactly this process of change and evolution that has brought me to Oxford today. The OED, the world’s largest dictionary and the definitive guide to the meaning and history of more than half a million English words and phrases, is one of the greatest works of scholarship ever produced. But although supplements containing new words have twice been published (and were consolidated into a second edition of the OED in 1989), the core content has not been revised since the original work was published, section by section, between 1884 and 1928. Now, Simpson and his team are engaged in the task of rewriting the entire dictionary - a vast endeavour that could take 20 years to complete.

Why does it need rewriting? Unlike conventional dictionaries that only cover the present-day senses of words, the OED is also a historical work, tracing the way words have been used over the centuries, and illustrating the uses with quotations. Over the years, as more historical sources have come to light, lexicographers have built up a much more detailed picture of how individual words developed, making the entries for those words long overdue for updating.

More obviously, the dictionary needs to be expanded to include the new words and terms entering the language. Since the OED is now online, this can be done continually, although, in practice, the updating takes place once a quarter, accompanied by a bulletin announcing the changes. Among the new words introduced in the last update, in December 2006, were “ladyboy”, “Talibanisation” and “webcast”, while new subentries under existing words included “extraordinary rendition”, “work-life balance” and “Vietnamese pot-bellied pig”.

I want to know why some new terms such as “permatan”, “shoulder surfing” and “duvet day” have made it into the OED, while others, such as “podcast”, “happy-slapping” and “helicopter parenting”, have not. But first, I get a quick introduction to the OED’s origins from archivist Beverley Hunt.

Walking me around the small, in-house museum, with its rare documents and old photographs telling the story of the OED’s creation, Hunt explains how the idea for the dictionary arose in 1857 when members of The Philological Society of London - still a thriving organisation - decided the world needed a complete historical dictionary of English from Anglo-Saxon times onward.

In fact, several other English dictionaries already existed - most famously, Dr Johnson’s - but The Philological Society considered them inadequate. Johnson’s had only 43,000 words and, as Hunt points out, “a lot of his own prejudices went in there.” Often-cited examples include “Excise: a hateful tax levied upon commodities” and “Oats: a grain, which in England is generally given to horses, but in Scotland appears to support the people.”

Given the loftiness of The Philological Society’s aims, it is perhaps not surprising that the project was slow to proceed. But in 1879 the dictionary found a publisher in the Oxford University Press, which appointed James Murray, a Scots-born teacher at London’s Mill Hill School and a distinguished member of The Philological Society, as principal editor. Murray and his colleagues estimated it would take 10 years to produce a work including all the vocabulary from the Early Middle English period (about 1150) onwards, but when after five years they had only reached “ant”, they realised their schedule was in need of a rethink.

The OED was eventually published as a part-work in 128 sections, called fascicles, typically selling at 12 shillings and sixpence apiece. When finished and bound in 1928, it contained more than 400,000 words and phrases in 10 volumes, and had taken roughly 50 years to complete. Murray, having devoted much of his life to the project, sadly never saw the fruition of this extraordinary undertaking - he died in 1915.

How do you set about compiling a dictionary? Murray’s method was not very different from the one in use now. Since the OED’s mission was to cover the entire history of each English word and its meaning, from the earliest recorded use to the most recent, Murray needed examples - lots of them. And the way he gathered these was through a reading programme, asking people to read books and literature of all kinds, past and present, to collect examples of word uses and write them down on index cards that could be collated with other examples of the same word’s use.

Thousands of readers, mostly unpaid and many of them responding to a public appeal Murray made in 1879, provided the OED with millions of quotations from written English. One of the more useful and prolific volunteers was W.C. Minor, a US surgeon whom Murray later discovered was a killer incarcerated in Broadmoor Asylum for the Criminally Insane in Crowthorne, Berkshire. The strange story of Minor’s contribution to the creation of the OED was told in Simon Winchester’s bestseller The Surgeon of Crowthorne: A Tale of Murder, Madness and the Love of Words, published in 1999.

Today, the OED’s incoming quotations are collated on an electronic database, yet the reading programme remains essentially the same. One part of it reaches back in time, going beyond the original dictionary’s main sources in the traditional canon - Shakespeare, Milton, Sir Walter Scott and the rest - to less well-known works, women’s writing and non-literary texts such as wills, diaries and letters. The other part of the programme focuses on modern-day English and new words, examining the way language is being used in all documented forms - not just books, newspapers and magazines, but scientific journals, song lyrics, television scripts, menus, advertisements and even the Yellow Pages.

These days most of the reading is done by paid professionals spread around Britain, North America and the rest of the English-speaking world. But anyone can send in a contribution if they spot something they think should be in the dictionary. Indeed, the BBC is this year planning to screen a second series of its Balderdash & Piffle programmes in which the OED invites viewers to help track down earlier uses of certain words and phrases than those already cited. Many of these words and phrases are slang or colloquial terms, which can be hard to trace in standard texts; for example, the OED is looking for verifiable evidence of the use of the word “plonker” before 1966, of “kinky” before 1959 and of “loo” before 1940.

Returning to John Simpson’s desk, I ask what sort of people write in to the OED with suggestions for dictionary entries. “Oh, very nice people,” he says in a scary sort of way, making me wonder if they are a bunch of latter-day W.C. Minors. But, he adds hastily, many of them make invaluable contributions - for example, working their way through 18th-century texts that the OED’s own staff would never have enough time to examine.

“They are people who want to see the history and documentation of the language improved, and see that it can be,” Simpson says. “They like contributing to an international effort. That’s some of the motivation that the editorial staff have here: the feeling that they’re contributing to one of the major research projects in Britain today - in the humanities, anyway.”

Simpson takes me to meet Bernadette Paton, an associate editor. She is working on the revision side of the dictionary, reaching back into the past and updating the entries for existing words to take account of our much improved knowledge of their evolution.

Paton explains how digital technology has transformed her task. Exhaustive as the original dictionary may have been, it was only practically possible for its contributors to read a small fraction of the material that was available at the time, because there was a limit to how many libraries and collections they could visit, and how much time they could spend reading the books to which they gained access. But now, without leaving her desk, Paton can conjure up this literature on a computer screen from vast online databases that most of us have never heard of.

One example of these databases is Early English Books Online, containing the text of about 100,000 books from the mid 1470s, when the first book was printed, to about 1700. The online Middle English Dictionary, a product of the University of Michigan, goes back further still, tracking the use of English in manuscripts dating from just after the Norman conquest to the introduction of the printing press. A US website called Newspaperarchive.com carries the contents of 2,463 American newspaper titles dating back to 1759, representing more than 58.1 million newspaper pages.

In most cases such databases not only carry scanned facsimiles of the texts they collect, but have digitised all the words on the pages, making it possible to search the entire database for individual words or phrases and find their earliest recorded uses. Paton tells how it was just such a keyword search, in this case on a database of historical books and journals called the Making of America run by New York’s Cornell University, that revealed how the term “motor car” originally meant a railway vehicle that travelled under its own steam, a much earlier meaning than the one given in the original OED.

Boorishly, I ask why it matters. “It’s not just about the language. It’s about tracking history through the language,” Paton enthuses. “Take the early-17th century, for example, Shakespeare’s time. This has always been seen as the period of the great flowering of the English language, of the coinage of many words. But our historical work is showing that many of the words thought to have come into use in that period actually go back much earlier. So our picture of the evolution of the language is being changed.”

Why did everyone think English blossomed in the 17th century? “Because that was when printing really took off, so there were a lot more printed texts from that period. Now, we’ve got resources at our fingertips like the Middle English Dictionary - a great historical record of the language of the medieval period [drawing on] documents such as wills, inventories and all sorts of non-literary material that the original OED editors just didn’t have access to. They only had access to the printed material, so that skewed the evidence in favour of the Early Modern period.”

Since Paton’s knowledge of the language evidently goes back a long way, I ask her which English words score highest for longevity. She says they tend to be simple, high-frequency, core vocabulary words: the pronouns, articles, number words, prepositions and conjunctions, plus core function words such as “walk” and “run”. (Later, Simpson sends me a list that includes “be”, “can”, “do”, “make”, “may” and “will”; “to think”, “to eat”, “to see”, “to sleep”; “father”, “mother”, “brother”, “sister”, and “church”, “house”, “wall”, “grass”, “cow”, “horse” and “sheep”.)

If you think about it, Paton says, many new words are just old words that have acquired new meanings. Take “screen”, for example, which is a very old word that originally meant something to ward off a draught or the heat of a fire, but has acquired many new uses: in motoring (windscreen), in the cinema (screenplay), in television (the small screen) and in computers (screen saver), to name a few. “And that’s what we’re dealing with as we write new definitions, quite often: we’re just dealing with an old word that has acquired a new application.”

Since the talk has switched to new words and new meanings, I move on to meet Graeme Diamond, a principal editor at the OED, and ask him what it takes to get a new word into the dictionary. I already know the basics: once in the OED, a word is never deleted, so the editors need to be sure it has caught on and become an established part of the language. That means words used only for a short time or by a small number of people are ruled out.

Diamond says it really comes down to one thing: evidence. “We want to see general usage of the word, and we want to see it used unselfconsciously, by which I mean, with the expectation of being understood. A lot of brand-new words nearly always appear with an explanation of what they mean straight afterwards, which is not the kind of usage we’re interested in - because that’s not a part of the English language, that’s a word made up of English letters that a writer is still having to explain.”

A great example of a word that could be heading for the OED but is not there yet, says Diamond, is “Wag”, or “Wags”, as in wives and girlfriends. Its progress, he says, has been remarkable. “It has gone from being introduced last summer in the World Cup to the point where, yesterday, I was reading an article in a national newspaper that completely unproblematically referred to ‘darts Wags’ in the context of [an international] darts tournament. A year ago, for a writer in a national newspaper to use the phrase ‘darts Wags’, and expect to be understood, would have been unthinkable.”

Is it true that a word needs to have been in use for 10 years before it can make it into the OED? It is only a rule of thumb, Diamond says; the editors exercise their judgment. “But underlying the 10-year rule of thumb is something that points to how philosophically different the OED is from other dictionaries, and that is our responsibility not just to tell you what a word means but to give you a historical perspective on it. That’s the reason we won’t be publishing ‘Wag’ any time soon, because we want to see what happens to it.”

Still, says Diamond, if a new word does become commonly used and understood in a wide enough context, if its meaning has stabilised and if the word shows no signs of fading away, the 10-year rule may be bent - as it was for “chav”, which was published in the online OED in 2006, only eight years after its first verifiable use in 1998.

Surely, though, people were using “chav” colloquially before 1998? Very likely they were, Diamond says, but unless someone can provide hard, documentary evidence, “we can’t do anything about it, because one of our principles is that everything we cite must be verifiable. We can’t just have someone saying: ‘I was using this in 1992.’” The spoken word is no use if it goes unrecorded - although, interestingly, an internet quotation may be usable if it can be printed out and kept in the archives.

I am still thinking about new words when I return once more to Simpson’s desk. How do they get invented? What determines whether or not they catch on? And are more of them being generated than ever before?

Taking the last question first, Simpson says it is not so much that words are being generated at a greater rate, but that there are more English-speaking people than, say, 100 years ago, and they are spread more widely around the world. “The more people there are speaking English in different places, the more change is going to come about” - especially in today’s multimedia age, he adds, which can spread a new word around the planet in a heartbeat.

New words pass into the language for fairly obvious reasons: they are attractive and fill a gap. But curiously, Simpson says, entirely new words are rare. “Word formation is based largely on familiarity. You are much more likely to accept a new term if you can have a stab at what it means, so most new terms reuse old ones, or bits of them.”

Thus, one of the commonest ways of forming new words is conversion, or changing one part of speech to another. Much as language purists may deplore it, the list of nouns that have turned into verbs is almost endless, as in “access”, “sidetrack”, “mouth”, “signal”, “text”, “screen” and so on. Another big source of new words (and purist fury) is semantic shift, where an existing word acquires a new meaning. “Gay” is an obvious example - today it rarely means joyous or lighthearted; “computer” used to mean a person who computes; and a “board” was once just a flat piece of wood, as used in making a table, but became the governing body of a company because that was what the directors sat around.

Next, we have prefixation, as in “declutter”, “unfamiliar” and “postmodern”; then suffixation, as in “happiness”, “computerese” and “radicalise”; and, of course, the ever-popular compounding, as in “cupboard”, “binge-eating” and “snowboarding”.

Importing from another language is perhaps the other most obvious source of new words. The English language is crammed with foreign borrowings, such as “curry” (from Tamil), “automobile” (from French) and “sudoku” (from Japanese).

As for wholly invented words, they are so rare that it is hard to think of many. Examples that Simpson comes up with are “quark”, now a scientific term but lifted from James Joyce’s Finnegans Wake; “galumph”, which comes from Lewis Carroll’s Through the Looking Glass; and “nylon”, which seems to have been plucked out of the air by DuPont scientists in the late 1930s.

So, I ask Simpson, how come “permatan”, “shoulder surfing” and “duvet day” are in the dictionary, while “podcast”, “happy-slapping” and “helicopter parenting” are not? The last three, he says, have so far failed the test: “podcast” is barely three years old, “happy-slapping” (hitting someone while an accomplice records it on a camera phone) is not only new but is showing signs of fading, and “helicopter parenting” (hovering obsessively over your children’s lives) still requires that telltale explanation. But in the case of “permatan”, “shoulder surfing” and “duvet day” - well, he says, many seemingly new and trendy terms are older than you think.

He is right: on looking up “permatan”, a word I thought had been around only two or three years, I am surprised to find its first recorded use was in the Washington Post in 1984. Similarly, “shoulder surfing” - stealing people’s passwords or pin codes by surreptitiously watching what they are doing on-screen - appeared in Computer Decisions magazine as long ago as 1985. And the earliest recorded evidence of “duvet day” dates back to 1996 when it was used (yes!) in the Financial Times.

For the price of an annual subscription to the online OED - ₤195 - you could almost make a parlour game of it, guessing the earliest recorded use of words or terms. “Metrosexual”? Described at length by The Independent in 1994. “Nethead”? Appeared in a Usenet message on a computer network in 1984. “Gastroporn”? Identified this year by the advertising agency JWT Worldwide as “one of 10 now phrases that will shape our lives in 2007” - but appeared in the New York Review of Books in, wait for it, 1977.

Enough of games; Simpson has a dictionary to edit, and I am getting in his way. Before I go, though, there is something I have to ask him. Although new words are being added to the dictionary regardless of where they fall in the alphabet, I notice that the revision people, who are working through the alphabet letter by letter, have reached the letter Q. Does that mean the epic task is nearly complete?

Not exactly. They cheated, apparently, by starting at M. The reason, Simpson explains, is that the early parts of the original OED are the weakest; after all, the editors were relatively inexperienced when they wrote them, and at that stage much less reading had been done. It seemed foolhardy to start with the letters that would need the most work, he says, “so we just said we’d start somewhere in the middle with a letter that had a good mix of Romance and Germanic words and M seemed to fit the bill.”

Just to complicate matters further, they are not going to work their way through to Z, then go back and do the first half of the dictionary. Instead, they are flipping back to the beginning of the alphabet once they have finished R, then working their way through to the end missing out the letters they have already done.

Why? Some scholarly dictionaries that the OED draws upon have not yet reached the end of the alphabet, Simpson says, and the OED’s staff will be experienced enough for the harder letters by the time they have finished R. Besides, he adds archly, “we’ve got to finish at Z, haven’t we?”

But whichever letter comes last, the job of revising the OED will never be complete. Now the dictionary is an online product, changes to its content can be made all the time instead of having to await the publication of a new edition once or twice a century. Indeed, once sales of the second edition dry up, the OED may never appear in print again.

Perhaps that is why Simpson seems unworried by the idea that he may have retired by the time the revision reaches the final entry (”Zyrian - a member of the Komi people of northern central USSR”). His job of turning the OED into a living dictionary, forever evolving and moving with the times, is already done.

“It will continually need updating and maintaining,” he says. “There’ll be new words, of course, and there’ll be more research that one can do on older words, more databases will become available, new editions of old texts will be published... it’s a never-ending task.”