You can say that again! A day in the life of an Actor-Phonetician
At the end of last year, a mammoth update meant that OED subscribers can hear words spoken aloud for the first time, in both British and American accents. Little triangles have appeared next to the transcriptions, and can be clicked to hear the word. It’s now quicker and easier than ever before to find out how a word is pronounced.
Now a pronunciation editor, I had the pleasure of being one of OED’s speakers periodically for a couple of years (I’m the British voice of entries including industrial revolution and pisteology). Our speakers have to be fluent readers of phonetic transcription, which makes us a little different from most voice actors and why our regular speakers are sometimes described as ‘actor-phoneticians’. Phonetic accuracy is vital – sometimes OED needs to show two, three, four or more accepted pronunciations of the same word, and these can be quite subtle differences (to date, the greatest number of British pronunciations listed for one word is sixteen, for endurance, while impasse has eleven – not all of these have published recordings yet). But it is not enough to simply have meticulous attention to detail – each recording also has to sound natural rather than stilted or robotic. Each actor-phonetician agrees his or her own working pattern with OED’s Head of Pronunciations and audio engineer, but here’s an outline one of my typical days at the office.
A typical day at the office
Up by 6am, be sure of a good breakfast but out the door by 6.30. Train at 6.50, into Oxford just after 7.30 and a ten minute jog along the canal to an unassuming town house in Jericho, Oxford, backing onto the main OUP campus. These are the offices of the English Language Teaching Dictionaries division, but in a room down in the basement, surrounded by network servers and old editions of various OUP titles, sits a sound booth. This is no glamorous Abbey Road experience – just a soundproof unit the size of a toilet cubicle, in a fairly dark room aside from a standing lamp, a desk light and a little window up to the street. Despite this, it’s a lovely place to work – it feels like a different world. Next to the booth sits Gary, our audio engineer, whose first job is to print the pages of words to be recorded that day.
After a quick catch-up and a flick through the day’s set, it’s into the booth, hanging the pages on the wall behind the microphone. Each page has around 50 words laid out in a table, with the spelling on the left and phonetic transcription on the right. Most actor-phoneticians pay little attention to the left-hand column, but it’s useful for spotting little typos that have slipped through the net. In pronouncing each word, each sound needs to be accurately but naturally articulated, without rushing or exaggerating it. Moreover, each word should sound like its standalone citation form, rather than as if being read from a shopping list; sounding natural while doing this is difficult when the words have no context. The nature of recording also sometimes means doing things that feel unnatural in order for them to be clear on the audio, such as the way you release a p at the end of a word, or the way you transition between the r and l in words like choral. We don’t nail every one first time – just like experienced gymnasts, articulatory gymnasts trip and fall sometimes! Often it can be a simple misreading, but it can also be due to a sudden change in pattern. It feels good to get into a rhythm of similar-length words with similar sounds and similar stress patterns, but that can also make it easier to miss subtle differences or trip up when it changes.
After recording a batch of words, it’s time to review. Each batch is recorded as one sound file, so Gary splits it up into separate files. We both listen carefully and check that the recordings match the transcriptions. If there’s a slip, or the audio isn’t clear enough, we re-record it; if multiple attempts at a word were made during recording, we pick the best, or decide to record another. Occasionally there’s a typo or a query about a transcription and these are flagged for review by a pronunciation editor. Reviewing time is also a chance to rest the voice and rehydrate from being in the warm booth.
This loop of record-review continues until lunchtime, with a much-needed hour off then resuming until 3.45 before heading home for a rest! It’s up to the actor-phonetician how many words to record before each review – when I started, I’d only get through 50 words at a time, but ended up recording blocks of between 250 and 1000 words between reviews. This is why I found a good breakfast was important – both for stamina and to prevent stomach rumbles from interrupting longer words like infundibuliform or uroporphyrinogen. The concentration required for reviewing also limits the block lengths – an experiment with a 1500-word recording block seemed fine until we found ourselves having to listen carefully to a word list over an hour long (which took 2-3 times that that to review).
A typical full day would mean getting through between 1500 and 2500 words, dependent on the number of typos, queries, and the complexity and similarity of the words. To date, the record for a single day stands at 3000 recorded words, only possible because of a long run of words beginning with over- which had very similar stress patterns! To many people it sounds like a dull job, but to a phonetician it’s a thoroughly enjoyable articulatory workout and my competitive streak kept me striving to nail more of them first time around.
Coming to a word near you…
It’s an ongoing process, but every non-obsolete word in OED will eventually be recorded. With OED’s expanded coverage of World Englishes, it applies across varieties, too – we’ve had Scottish, Irish, Canadian, Australian, New Zealand, South African, Hong Kong, and Philippine English speakers… and we’re recruiting more speakers as we add varieties.
So why do we still need the symbols?
There are lots of reasons why the phonetic transcriptions are still needed. The transcriptions are used by our speakers to make recordings, but they also benefit users who are unable to access the spoken pronunciations (because of hearing difficulties, or they have no headphones/speakers). Even more importantly, the recordings reflect each speaker’s individual speech features and it may not be clear how far you could deviate from his or her pronunciation and still be recognizably saying the same word. The transcriptions are broader, and when properly interpreted allow for a range of subtle differences.