Word for word

by Emma Hardman

'Letters are signs of things, symbols of words,
whose power is so great that without a voice
they speak to us the words of the absent;
for they introduce words by the eye, not by the ear.'
– Isidore of Seville, Etymologiae, c. 600


WHEN I TELL people I work for Hansard[i], the conversation generally goes like this: Ahh, they say, So, you... They raise their hands and tinkle their fingers. They are miming stenography or Computer Assisted Transcription (CAT).

No, I say. CAT writers – and steno writers before them, and pen-and-paper shorthand writers before that – take down speech at the speed it is spoken, in real time. The stenotype machine – which is like a shorthand typewriter – replaced paper-and-pencil shorthand in the mid-to-late 1970s in Hansard. In the 1980s steno became CAT. The difference between a CAT machine and a stenotype machine is similar to the difference between a manual typewriter and a word processor: same keys, same method, same shorthand language – different technology. Steno creates shorthand in ink on paper; CAT produces shorthand on a screen, which is changed by software into words. Steno or CAT is what you see in American courtroom dramas where a woman sits between the judge and the lawyers, using what looks like a mini typewriter and, if asked by the judge, reads back what has just been said: I did it, your Honour. I killed him. There are no stenotype machines used in federal Hansard any more, and only a few people who do CAT.

Oh, so you must know shorthand?

No. Although many Hansard editors do. Until the 1980s, Hansard reporters would take down everything that was said, in shorthand, with a pencil and paper. They would take their transcription back to an office, where they would dictate it to a typist, editing the report as they read it out. In those days, it was a lot less verbatim than it is today. 'Tidying things up' was an important aspect of Hansard' s brief – rewriting whole paragraphs, restructuring speeches, clarifying arguments, changing vocabulary and generally 'nicing' it up. The typed pages would then be edited and proofread again before being printed.

You must be a fast typist? They are ready to be impressed.

I figure they' ve earned it for maintaining their curiosity. You see, I say, Hansard is not a verbatim account. It is not a direct transcript. We transcribe what they say and we edit it, but it is more than that.

Now the look of shock and suspicion – which is encouraging: it shows that people have a sense of ownership of the words of the people they have elected to govern. They want direct access, transparency, between the people and their government, and this is something they expect as their democratic right. And I wholeheartedly agree.

When I tell them that it takes one of us an hour and thirty minutes to transcribe seven minutes and thirty seconds of parliamentary proceedings – together, known as a 'turn' – they are appalled. I can see the unasked question in their eyes: Can' t they hire people who can type faster than that? They are restrained enough to say only: But what on earth are you doing?

I try to explain that we are making the written record more accurate.

More accurate than what was actually said?

No, 'more accurate' as in reflecting the meaning of what was said more accurately than if we just wrote down everything they said word for word.

So, you take out the ums and ahs?

Yes, but it is more than that.

I try to describe how we transform spoken language into written language and how this is not the same as verbatim transcription. I talk about how people don' t always say what they mean, how speech consists of not just words but also facial expression, pauses, intonation, gesture, context, pace, volume; writing is just ink on paper. Sounds go out of a mouth and Hansard transforms the meanings such that they can enter someone' s eyes. The momentary becomes enduring; time becomes matter. It is an act of translation.

I am still facing a dubious look. Whatever gets you to sleep at night, I can see in their eyes.

Hansard is a 'rational verbatim' report. This enigmatic term evolved from Erskine May' s A Practical Treatise on the Law, Privileges, Proceedings and Usage of Parliament (1844). The modern Australian Commonwealth Hansard mission statement, derived from this Westminster antique, is: 'To provide an accurate, substantially verbatim account of the proceedings of the parliament and its committees which, while usually correcting obvious mistakes, neither adds to nor detracts from the meaning of the speech or the illustration of the argument.'

It sounds clear enough. But how do you know what substantial is, in substantially verbatim? How often is usually, in usually correcting obvious mistakes? How do you know if a mistake is obvious or not? And what about adding or detracting from the meaning of the speech or the illustration of the argument? By what standard?

Writing is 'the graphic counterpart of speech, the fixing of spoken language in a permanent or semi permanent form' .[ii] The first words for 'reading' in Greek and Hebrew also meant 'to convince by argument or rhetoric, to call out, to recite' .[iii] Up until quite recently – the Middle Ages – reading was speech. No one read in their head, as we do now. Even when alone, reading and writing were spoken. Before the ninth century, scriptoria in Western Europe were very noisy places. Everyone was there to read and to write, and both of these activities by definition meant speaking the words aloud.

The rise of Carolingian minuscule – a script that introduced punctuation, capital letters and spaces between words in Western Europe – meant that the reader did not have to work out where each word and sentence began and ended. Where there were three modes – sight, hearing, speech – there became just one: sight.

By the thirteenth century, silent reading was as entrenched as reading aloud had once been.[iv] The new practice of silent reading sent rules to the other extreme. In the scriptoria, all spoken language was banned and scribes were required to communicate with each other in sign language while working. Scriptoria fell silent.

Stephen King says that 'writing is telepathy' ,[v] and the British writer, translator and linguist Anthony Burgess, author of A Clockwork Orange (1962), observed: 'Thoughts, desires, appetites, orders have to be conveyed from one brain to another, and they cannot easily be conveyed directly. Only with telepathy do we find mind speaking straight to mind without the intermediacy of signs...The vast majority of sentient beings – men, women, cats, dogs, bees, horses – have to rely on signals, symbols of what they feel and think and want.' [vi]


TODAY IN HANSARD the majority of editors use voice-recognition software (VR). The editor sits in the chamber only to 'log' what is going on: a kind of minute-taking. They are not typing everything that is said; it is all being recorded by the broadcasting department. The editors are in the chamber to write down who is speaking when. And you won' t find many of those colourful interjections that are flung across the chamber in the transcript, because they only go in if they are responded to by the person who has the floor. After seven minutes and thirty seconds, the next editor on the roster comes in and the first editor goes back to their office, puts on their headphones and calls up the audio of the proceedings they have just witnessed. As the sound goes in their ears they speak what they are hearing into their microphone, and VR produces those words on the screen. We call this 'voicing' . When you voice, it works best if you speak fast, because the technology uses context: the more word groups it has at once, the better its 'guesses' about what you said and what you meant.

'Words' don' t actually exist in spoken language. Although we might think we say the warmth there was, what we actually say is closer to the warmpth air was, which VR has to make a stab at and might come up with the warm there was or the war there was or the war path was. This data sounds the same as this starter or this Tartar. We rely heavily on intonation, non-verbal signals and contexts to decipher the speaker' s meaning.

Language is not simply a means of communication; it is also a crucial part of the entire process of cognition. We use words to designate objects and their location in space. Through grammatical constructions we express relationships and ideas. Language is fundamental to perception, memory, thinking and behaviour.

Our brains devote a vast amount of space and energy to language functions. There is reading, writing, speaking, listening, hearing, body language, gesture, facial expression. Each of these involves motor functions as well as cognitive function: you move your jaw, hands, face, lips and tongue to speak; you move your hand, arm, head and eyes to write; and you move your eyes to read.

We tend to think of language as one thing. You can say (or sign) the word dog, you can hear the word dog, you can write the word dog and you can read the word dog. It still means the same thing. But reading, writing, speaking and hearing are all different processes, physically and neurologically. When a conscious human subject is scanned using positron emission tomography (PET) and asked to perform similar but subtly different language tasks, different areas of the brain light up, depending on whether the subject is hearing words, seeing words, speaking words, even generating verbs. [vii]

The same part of the brain lights up when a subject is speaking, no matter what the language. (This is the same part of the brain that lights up when the subject is a sign-language speaker, whichever the sign language.) When reading or writing, it is another part of the brain – again, it is the same, no matter what language.

The alphabet is no more than three thousand years old, but speech is as old as humans. We tend to think of writing and reading as natural upshots of speech; however, the relationship between written language and spoken language is arbitrary. The English alphabet has five symbols for vowels: a, e, i, o, u. But English has twenty-three vowel sounds: fifteen single sounds and the rest diphthongs (double vowel sounds, as in eye) and triphthongs (as in our). The acrobatics performed by the brain to hear these sounds and decode them into bits of meaning using the Latin alphabet (a, b, c, d, e...), or vice versa, to use only five symbols to cover twenty-three vowel sounds, is an incredible feat.


LIKE ANY OTHER kind of translating, Hansard editing is problem-solving, a constant balancing of many rules. The main rules we have to deal with are: a Hansard transcript should be clear, readable and grammatically correct; it should accurately record what the speaker said; and it should accurately record what the speaker intended.

These three rules are often mutually exclusive. And add to this roughly a thousand pages of others – relating to punctuation, grammar, usage, spelling, hyphenation, capitalisation – contained in the Hansard style guide, the Hansard editing guide and the Hansard form guides, not to mention our other staple, the Macquarie Dictionary, all of which contain rules that are often in conflict with each other.

In the passage from a speaker' s brain to the brain of a reader of Hansard, there are many channels that meaning has to pass through. The editor has to find the most direct and clear path through these channels, changing what the speaker said into something that has the same meaning but consists of marks on a screen – letters and numbers and symbols in space; no longer sound waves passing through air. It is a fabulously circuitous route.

The first hurdle is the speaker' s intention: what the speaker thinks or means, before they even move their mouth, those pre-verbal firings of electricity in the brain.

Then there is the speaker' s attempt at a transformation of those electrical currents and snapping synapses into words – still not spoken. We can never reliably recreate our cerebral intentions: there is always a gap between intention and action. This gap, the gap between thought and speech, between meaning and words – the gap we fill with language – is tiny, and huge. It is a black hole that all words fall into, and it represents the entire endeavour of communication, of writing and reading and conversing with each other. This gap is the attempt at telepathy that Stephen King and Anthony Burgess speak of. To say or write what you feel and experience and see and hear, to communicate what you mean, is a subject that philosophers have spent lifetimes trying to untangle.

After this, there is what the speaker actually says. The speaker might say We did not do that, when they meant to say We did do that. This may seem like a big error – to say the opposite of what you meant to say – but it is surprisingly common. A verbatim rendering of this would be We did not do that, but a Hansard editor will have taken into consideration all of the surrounding utterances, the context, the tone, the political alliances of the speaker, the body language (perhaps the speaker was distracted at that moment, by someone entering the chamber), and deduced that the speaker really meant to say the opposite of what they said.

Then there is what the editor hears. It took me a while to realise that it is worth listening to that confusing phrase, that swallowed word, just one more time, even though you' re sure you won' t be able to hear any more than you did the first five times because, miraculously, on the sixth listen something has popped in your brain, something has changed, and it is clear what is being said. Everything is the same, but your brain has processed and focused on every detail until it finally homes in on the one thing you were not able to understand. They are saying when you look at this data not when you look at the starter. They are saying refugia not re the future. And this might shed new light on the rest of the sentence or paragraph.

Then there is what VR hears the editor say – and this is easily fixed, but makes for some fun: catch lick priests or deigned for Catholic priests ordained; who were turpentine indisposition for who were there at the time in this position; 2000 and Sikhs for 2006; and Collette Ding and burying parent emissions for collecting and burying their emissions.


IT IS AT first counter-intuitive that a verbatim account of the proceedings of parliament could be less correct than a 'rational verbatim' translation. But the verbatim translation of 'verbatim' cannot be done in one word; you need three: 'word for word' . We can say the most confusing things, yet the listener extracts the right meaning. Every moment, our brains glide over misspeakings, repetitions, stumblings, mumblings and a general inability to express ourselves clearly, to put our thoughts into speech. We are also interpreting winces, pauses (does the speaker pause because they have lost their train of thought, because they are adding emphasis, or because they have just realised that they are reading from the wrong sheet of paper?), throat clearings, gestures, the expression in the voice, the emphases. It is impressive how accurate the human brain is when it tries to extract meaning from such confusing data.

Here is an average piece of spoken language written verbatim:

In terms, of, um, sorry, just to, I mean, in terms of this very first one, you can really see here. Um, Peter, would you, thank you. In terms of this getting, getting out of canoe, it initially didn' t, at first it really strikes me that really bush stores what it seems to be is it seems to be quite a powerful idea in that sense in as much as it would seem too if it' s really very much improving one aspect of those communities being a store that really really reduces the two I guess but it also as well and it would seem is providing an example of a business prize beyond what the really small list of examples of the jobs you have got in those communities right at the very present moment which are providing a really a role model. What it is is that. Do you agree with that?

Gibberish, right?

Here is a Hansard-style edited version, the result of close attention to context of topic, of issues and of intonation; to pauses, pace, emphasis, common sense, what was said before and after, who the speaker was and what was happening in the room (a PowerPoint slide was being shown):

In terms of this first graph – getting out of the 'goo' – it initially strikes me that Bush Stores seems to be quite a powerful idea in the sense that, if it is improving one aspect of those communities, being the store, that reduces the 'goo' . But it also would seem to be providing an example of a business enterprise beyond the small list of examples of jobs which are providing role models in those communities at the moment. Do you agree with that?

This is not elegant prose. But it makes much more sense. It is an effective translation from one mode to another, a written representation of what was said and what occurred.

A Hansard editor must have a passion for the nuances of words, phrasing, punctuation, usage, grammar, vocabulary, syntax – and a broad general knowledge. You must know (or know where to look to check) that the bee' s knees refers to the knees of one bee, not the knees of many bees (which would be bees' knees). You receive your just deserts not your just desserts (it comes from the verb 'deserve' ). It is grisly remains not grizzly remains (unless all the other bears have left the campsite and only one grizzly remains). And if the speaker says It' s not rocket surgery, the editor must decide if it should be rendered: It' s not rocket science or It' s not brain surgery, or if it should be left as is.


HANSARD TAKES OUT false starts, redundancies, verbal tics and unnecessary repetitions. We get rid of all the ums and ahs, the ahems and errrs. We get rid of 99 per cent of the Mr Speakers and Mr Presidents that bookend so many utterances in the House of Representatives and the Senate. Other verbal tics that are candidates for deletion are very, really, actually, certainly, however, surely, of course, so, therefore, and, but, because and I think.

Hansard editors routinely put in things that were never said and leave out a whole bunch of stuff that was said. Most of this stuff is 'form' . Form is procedural text – replacing what was said with standard descriptions. For example, when we hear something like this:

President: Is leave granted to move that? Leave is granted. You can now move it Senator Nettle. You move the postponement to the next day of sitting? That' s 810 in your name?

Senator X: That' s right.

President: The question is that that motion be agreed to. Those of that opinion say aye, those against no. I think the ayes have it.

The transcript will read Question agreed to – a phrase that wasn' t even said. This kind of translation is crucial in representing accurately what has happened in parliament: a bill has been passed or a vote has been taken.


ONE OF THE things that sucks up a Hansard editor' s time in the desperate dash to get their turn finished in an hour and thirty minutes is checking. We check spelling, punctuation and grammar. We check the names of organisations and departments, television stations and mining companies. We check the date the second reading debate is being resumed from. We check the spelling of Princess Leia, the Fonz and Jabba the Hutt.[viii] We check page references, committee titles, names of countries. We check Aboriginal language names, acronyms (there are thousands of these in regular use) and the names of the Auburn under-eights netball team.

We discover that seedoor is CEDAW: the Convention on the Elimination of All Forms of Discrimination against Women. We need to ascertain what the sugarloaf salient is, what a retention lease is, and what bitterns and vuvuzela mean, to know how to spell them and to verify that what we think we heard makes sense in the context. And we check names, even if they seem obvious. Invariably, the John Smith whose name you neglect to check is Jon Smythe.

We check quotes. And politicians quote a lot. They quote from newspaper reports from fifteen minutes ago to three hundred years ago. They quote from television shows, movies, radio interviews and songs – in June 2008, Mr Raguse, the then member for Forde, sang the first verse of Redgum' s 'I Was Only Nineteen' . They quote TS Eliot, Henry Lawson, Shakespeare, Emily Dickinson, Socrates and Ronald Reagan. They quote from Senate committee reports and from Hansard. They quote from legislation, letters and emails, international treaties, scientific studies, websites, annual reports and advertising material. And they quote from subsection two, paragraph two of the legislation and Hansard has to work out that what they meant was Section 146A(2)(b)(ii) of the Cross-Border Insolvency Bill 2008.


AS A LOVER of language I had always been aware that spoken and written language are different. Spoken language is looser; you can get away with more. You don' t have to think about spelling or punctuation when you speak. And most people are aware that body language plays a larger part in communication than we consciously perceive. But how profound the differences are was unexpected and fascinating. What I love about my job is being with the words and meanings in an intimate way; wrestling the spoken word into the written; translating one to the other; and being the closest thing there is to a telepathist, a medium between one brain and many other brains. The transcript that Hansard produces is not word-for-word. But nor is it a painting, a collage, an artistic interpretation of what went on. It is a clear, well-framed photograph.


[i] 'Hansard' is a generic term for transcripts of parliamentary proceedings. In Australian federal Hansard, that means pretty much everything that is said in the House of Representatives, the Senate, and Senate and parliamentary committees. In Australia, New Zealand, the UK and most other Commonwealth countries, transcripts (federal and state) are called 'Hansard' . 'Hansard' also describes the people who produce these documents – in Australian federal Hansard, a department of around fifty people – and the work they do. People who work at Hansard are often referred to in the second or third person as 'Hansard' : Hansard, are you ready? (at the beginning of a committee hearing); Hansard, did you get that? (in the chamber); and Thank you to Hansard. Luke Hansard was a compositor, and later, his son, Thomas Curson Hansard published the parliamentary debates of England from 1770 onwards. 'Hansard' appeared on the front of every report because that was the name of the printer.

[ii] Fischer, A History of Writing (quoted from David Diringer, Writing, 1962), p. 11.

[iii] Fischer, A History of Reading, p. 11.

[iv] Fischer, A History of Reading, p. 164.

[v] King, On Writing, p. 95

[vi] Burgess, p. 8.

[vii] Greenfield, p. 37.

[vii] Yes, these three characters were mentioned in parliament.

From Griffith REVIEW Edition 31: Ways of Seeing © Copyright Griffith University & the author.