Back to: Gadget patrol: the new iPad | Forward to: Pirate LOSS? An alternative ...

A writing experiment I plan to try

Fiction is an art form, and like all art forms it relies for much of its effect on a bunch of shared conventions that both the author and reader are familiar with. There's a difference between a bland description of a sequence of events, described in the passive voice from an omniscient viewpoint—as in, for example, a report by a formal body investigating an accident—and a dramatic presentation intended to induce a specific emotional reaction in the reader. And one of the key techniques we use in written fiction is dialog.

Next time you read a novel, try speaking some of the dialog aloud. Then analyse a conversation between real human beings (as opposed to imaginary ones) — preferably by recording one in the wild and then transcribing it directly to text. (NB: It helps if the conversationalists aren't self-conscious about being recorded.)

Human conversation doesn't resemble written dialog; it's messier, full of silences and throat-clearing and punctuation speed — stuff we mentally filter out, "er" and "like" or "fuck" (in Scottish — yes, up here it's frequently used as a punctuation noise, especially in city centres after mid-evening). Words and whole phrases go missing, replaced by hand-waving, face-pulling, and other gestures. Sentence structure goes awry, taking second place to the exigency of breath control and the random twists and turns of thought, almost as if — hey where did I put my — wait, not there, there — we don't think (or communicate) in formal grammatical sentences.

The truth (which we are carefully trained to ignore, from the very start of our reading age) is that casual speech doesn't follow the formal rules of grammar or the structure of rhetoric. And so narrative fiction, as an art form, has come up with conventions for representing speech. "I'm talking to you, dear reader," said Charlie, deploying quotation marks (the double apostrophe) to indicate the beginning and end of a dialog block within text. The dialog markers indicate that you are meant to interpret the enclosed text in a different manner from the surrounding narrative by using your imagination to colour in my intonation and accent and the pauses for breath that are missing from regular narrative and that you will acutely notice the absence of if you try to read this sentence aloud. (Narrative text marches to a different beat, that of the comma, and, optionally, the colon and semi-colon.)

Because dialog markers are hard to spot inside a paragraph of text we usually place each chunk of dialog on a separate line.

"Like this?"

"By Jove, I think you got it!"

There are other ways of indicating dialog, of course: using an em-dash as a lead-in is one way of doing so:

—ye ken whit am sayin'?

"The disadvantage of denoting dialog in this way is that there's no trailing end-of-dialog marker, so each phrase or fragment of speech has to stand by itself with no descriptive metainformation to inflect the reader's interpretation of the words or provide a commentary on the content," Charlie explained at tedious length. "It can probably be thought of as a somewhat less formal version of script-style dialog," he continued.

READER: No, you're pulling my leg!


(Irvine Welsh would probably agree with me.)

Anyway, let us tip-toe rapidly past the syntax markers for dialog and get back to the semantics of the form. Good dialog shouldn't need much he-said she-said commentary, other than the minimum required to indicate speaking order in multi-person conversations; characters in fiction should be distinctive enough that the reader can tell them apart from context. Nobody has the same natural speech pattern, after all: you can tell people apart in a conversation even if you're wearing a blindfold. We choose different words, speak at a different pace, maybe speak in shorter or longer multi-word fragments, have different accents, and different pitch registers.

In dialog embedded in narrative fiction we have a much more restricted range of tools. (Some languages are inflected and have written characters to indicate this; English, not so much, although perhaps there are exceptions? More than one of them! But isn't it a bad idea to overuse them!!!!?1?!!??) In general we have to stick to the speaker's word choice, phrasing, and diction (along with a few extra twiddles we can use to indicate dialect — apostrophes to indicate dropped H's, for example).

Getting the phrasing and rhythm of narrative fiction right can be a bit of a challenge. One technique that many authors swear by is to read their manuscript aloud, and ideally listen to a recording of their reading of it. This will certainly expose some infelicities (over-long sentences lacking speech-pauses, for example). But I tend to think it's a snare and an entrapment. A reasonably proficient reader will average 300-400 words per minute. However, we don't speak anything like that fast: a machine-gun-rapid delivery is 200-250 words per minute, and regular continuous speech is more like 150-200 words/minute. Indeed, my rule of thumb for readings of my work is that I can read aloud 5000 words in 30 minutes, or 9000-10,000 words in a whole hour — although if I read for a solid 60 minutes my throat would give out. That's a reading rate of around 180 words/minute, and because it takes twice as long to get through each page by voice and ear as it does to read it on paper, certain aspects of the spoken word presentation of a written work don't come across properly. In particular, a 200 word descriptive paragraph (half a paperback page) takes an entire minute to listen to, but only 30 seconds to read. The minute drags, even though the 30 seconds doesn't challenge the attention span of a typical reader. Consequently, optimizing one's written prose for reading aloud may result in paring it down too far and sacrificing detail on the altar of brevity.

Anyway. The subject of this blog entry is: an experiment.

The pen, the typewriter, and the word processor are tools best suited to the task of extruding lengthy chunks of narrative. But they don't naturally capture the cadence and rhythm of speech. I am wondering, however, if speech recognition software hasn't advanced far enough to do a creditable job ...

Speech recognition has been around for years, and generally it sucks. I know authors who, due to hand or wrist injuries, have resorted to using it. I even know of an ongoing series of science fiction novels in which the author switched abruptly to Dragon Dictate between one novel and the next (due to an accident which left both his wrists in plaster while facing a deadline). It's been said that the shift from typewriters to word processors in the mid-1980s brought a marked change in the quality and structure of novel manuscripts; switching from typing to speech leads to at least as significant a change in any author's work because they don't work the same way. Keyboarding errors and speech recognition errors are different: the recognizer will often insert a similar-sounding but incorrect word into the stream of text, while the keyboard-induced typo is relatively easy to spot (indeed, 90% of them are detected by spelling checker, although some — their/there and similar — are much harder to detect automatically). When you're using a broad vocabulary that incorporates made-up words, speech recognition doesn't work so well.

However, what brief experiments I've done suggest that speech recognition can capture the cadence of dialog quite well.

So in the near future I'm going to do an experiment in multi-mode writing: that is, I'm going to use a keyboard to type the narrative, but switch to speech recognition software for dialog in the same text. Hopefully what comes out will be a more realistic rhythm. And if not? I need to refine my theory of writing.



A really interesting experiment, I wouldn't be surprised if a change of tool produces a change in results.

A side note: As a non-native English speaker, I hate it when different dialects are indicated by horribly mangling grammar and spelling. I don't mind a few 'h's replaced by apostrophes, but I've found some texts that are butchered beyond recognition. I very much prefer '"This is the house", he said in a very thick Franco-German accent' to "Zis ith zee 'ouss".

The worst cases I've found: Phonetically-transcribed Scottish accent. Completely indecipherable to me.


The worst cases I've found: Phonetically-transcribed Scottish accent. Completely indecipherable to me.

You know that Scots is not-quite-but-nearly a different language, right? Not only the accent but the grammar and vocabulary differs significantly from English.

(Random example: a "wally dug". Dug = dog, but wally? Is vernacular Scots for porcelain.)


If you want unrealistic, try speaking to your friends like a politician delivering a speech. And repeat everything three times.


I though "Scottish" was closer to Old English?


I really want to try those new speech-recognition things, like Siri, or the Android versions, or just dictation. But there is one issue where they all fall flat abominably: Multiple languages.

See, I live in Switzerland, and I speak a Swiss German dialect which sounds a bit like German around 1600 with a lot of (mispronounced) French words in it. I also speak standard German, and obviously English. Oh, and Japanese (with my wife, so actually really common). As long as I can keep to a single language, most tools can (barely) follow. But that's impossible. I will at some point have to reference things like names of places (often German around here, but sometimes French), or names of people, or just products or websites or anything else that is not in the Oxford dictionary. It's really hard to imagine how an American would pronounce Migros (French) or Stämpfli (Swiss name) or Kabetsu (Japanese), and even if I get that right, the program will not recognize the word. So I can spell it, and I have to go with the intonation of the language I am supposedly dictating in. I might as well type.


I don't believe so, but I'm not an etymologist or linguist. IIRC Scots is a Germanic language, rather less influenced by Norman French than southern English (no Norman invasion), heavily influenced by Viking Norse.


So it seems a fun experiment to try but I bet it won't produce dialog that feels more natural to your readers. As you say, our ears do a bunch of filtering of spoken dialog, so that what arrives at our conscious attention is rather different from what was said. Our eyes do a very different filtering of what we read. So when we read your speech-recognized dialog, we aren't going to hear your speech---we're going to get all hung up on a bunch of um's and restarts that we would filter out with our ears. You're asking us to play an mp3 through a wav codec.

If you really want us to hear speech, you should jump right to the audiobook version of your story. Or, write down that speech as it should be spoken, then "decode it" by staging a play/reading to deliver it through people's ears.

As a non-native English speaker, I hate it when different dialects are indicated by horribly mangling grammar and spelling.

As a native speaker of English, I have to concur. Accents may add colour, but they don't help in successfully conveying meaning.

Hopefully Charlie won't go that far in pursuing realistic diction (then again, a mere matter of making no sense whatsoever hasn't stopped people in the past).


Well, Old English still has a lot of German in it, including some grammatical structure. Often butchered in the fantasy genre and Star Wars (Yoda) - "My horse I shall saddle and with my brothers ride" etc


An excellent plan Charlie! Good luck with it. For what it's worth, I've found that Siri understands my horrible Australian accent about 85% of the time, and Dragon Dictate about 75% of the time - though I haven't really done much in the way of training DD to my voice patterns aside from the default 300 words.

I have found myself wondering if George RR Martin has heard of these kind of tools, and whether that would help him reduce the time between novels... but that's incredibly selfish of me to think that.

(goes back into my box)


Charlie (or anyone else) do you have any references to the effect of wordprocessing on literature? The best I have found so far is some journalistic reports of work by Professor Matthew Kirschenbaum, but they don't go into any detail about his findings.


Let's hope not as far as Banks in "Feersum Enjin" (or whatever)


Really interesting post.

"Am I Free to Go?", a story of mine forthcoming from, was written as a voice piece, which is to say it was built to be read out loud. And I read it out loud to a live human being, a theater director, about 26 times. Doing that changed my whole idea not only of prose composition, but also of what editing should be.

I'm not sure speech recognition is the right tech for what you are doing. Speaking in to a machine in such a way that it transcribes faithfully what you say is very different than speaking to another person in the voice of a character while being the character.

I suggest being less machine-oriented when actually speaking. Letting Dragon do its job, but also having your voice recorder to hand so the you know what you actually said and how you said it.


Hmmmm. Interesting. I'll be eager to see how that one works.


I once had a newspaper column where readers would send me questions, and my answers were transcribed verbatim exactly, including all repetitions, nonsense and placeholder words, pauses (duration included), and hand gestures. It was eye-opening.

One of my favorite modern novels is William Gaddis's JR. It's a story of massive financial chicanery, set in the New York City of the 1970s, centered around a precocious tween boy who manages to create a business empire from his school's payphone. It's mainly told in the form of unattributed dialogue. Warning: it's huge.


Excellent idea, it would be interesting to see the typed conversation, verses the dictated one!

I enjoy reading scots language, as a native scots speaker. It has some wonderful imagery. However I doesn't always work and sometimes it feels like the author has gone too far. Putting too much scots into a piece of writing without any change to modern English, as would be expected depending on the situations the characters find themselves in.


Hmmm. I'll be interested in the results. I feel like when I write dialogue I'm in a different "mode" than plain speaking, but I don't think it's entirely due to the fact that I'm writing instead of speaking. Literature has its own cadences, as you note. So we're all going to want to know whether you can achieve a smoother speech style or are instead broken out of your writing groove every time you try. Maybe it just takes practice.


I'd be curious about regional styles in conversational errors.

For example, in the field we US Americans tend to repeat ourselves for emphasis (too much cheap screenwriting consumed?). We often get lost in the middle of an exposition, forgetting where we were intending to go with that and instead moving on to a topic that is important to us emotionally. It's unusual to drop a fact into a conversation; but if you only do it once there will be amnesty. Adults brag by saying something they are proud of and then pausing a moment for you to admire them. Agreement and affirmation-zwang make bobbleheads out of groups of us, especially women.

One reason I enjoy your novels is because the characters don't do these things.


"I even know of an ongoing series of science fiction novels in which the author switched abruptly to Dragon Dictate between one novel and the next (due to an accident which left both his wrists in plaster while facing a deadline)." @JohnBirmingham ? He was getting dry this afternoon trying to keep up his word count.


that sounds like a fascinating experiment. I'd love to hear/read how it turns out, though most of your dialogue is pretty realistic, given its context. Personally, I've always found it difficult to - as you say - include the pauses, inarticulate noises & ear rubbings etc that are a part of natural speech. But I wonder - isn't that the thing with writing, as opposed to film? The imagination as opposed to the visual? Just a thought. Keira.


As alluded to in #13, the fact you're aware you're speaking to a transcription tool may affect your speech patterns. We're actually very good at modifying our speech patterns for our audience, particularly if we've had a bit of practice. Speaking to what I know, teachers in the classroom are much more formal in their speech than they are in more casual circumstances, and even the difference between junior and senior classes has huge effects on vocabulary, sentence structure, and tone.

The experiment is interesting, but it might be worth trying to take one step further and collaborate with someone to role play conversations while being recorded. Conversations can do all sorts of fun things, including looping back on themselves or referencing shared experiences that may not be understood by an outsider, but are easily recognised as such. This builds credibility.

If you do go with that idea, video might even help to capture some of the non-verbal communications that could then be described in text, he said with a shrug. You'd then have to transcribe the video, or play it to the dictation device to do so automatically (which may work if you drop the speed by 10% or so from normal rapid fire conversation - maybe).


Just to confuse things further, there's Scots, and then there's Scottish English.

Scottish English is English. It includes a few Scots words, but basically it's no more different from RP than Yorkshire English or Welsh English. It's what we've spoken in the Highlands since we lost the Gaelic, since we never spoke Scots, and I think it became more common in the Lowlands due to Scots being discouraged in schools.


Or "wag at the wa' ". Believe it or not, this is actually a single adjective and noun pairing in Scots. It refers to a wall-hanging pendulum clock, usually with a striking action, and with an exposed pedulum hanging below the case which serves solely to house the face and action.


Another writer who uses speech recognition is Sir Terry Pratchett, who I believe has said that if he regained the ability to use a keyboard, he probably wouldn't.


Er ... wouldn't return to using a keyboard, not wouldn't continue to use speech recognition. That sentence was perfectly clear in my head...


Nope: David Weber.


I recall reading a book that used no quotation indicator at all. It was seriously annoying, since I wasn't used to that. Although I think that after a while you learn to recognize the differences between quotation paragraphs and narrative due to the style of language.


Hmm - that's a bloody interesting idea. As a screenwriter, most of what I write is dialogue. And I've been considering testing out Dragon Dictate for a while now - I hear very good things about it from various people in web content creation.

I may join you in this experiment on my next project. If anything, it should work even better for producing dialogue that's actually designed to be spoken.

Of course, it may also mean I need to spend a month or so training my speech-to-text software to recognise all my different voices for characters...


I find some of those Swiss German dialects utterly weird. Bernese German seemed to my (layman's) eyes to be an unholy mixture of German and Welsh. Baseldytsch less so.

The worst I've seen was Baslerdüütsch written using a cod-Cyrillic typeface for a Zeedel. Ow!


The one book I recall doing that was an English translation of The Day of the Dolphin.

I found it was a trick that got in the way of comprehension.

The worst cases I've found: Phonetically-transcribed Scottish accent. Completely indecipherable to me.

Something like Trainspotting? I found that reading that became a lot easier if I read it out loud (or at least did the equivalent in my head). It helped to realise that "bairn" is similar to the Frisian "bern". (Frisian is the language spoken by a part of the people of Friesland, a northerd Dutch province. It is a real language, not a dialect, although the sentence structure seems remarkably similar to Dutch.)

Let's hope not as far as Banks in "Feersum Enjin" (or whatever)

Gods, I hope not. That might have been on a par with his otherwise good books if it weren't dragged down by the entire subplot being written in some sort of future textspeak-slash-phonetics.

Random thing on the topic of speech patterns being context-sensitive: I've seen a few people consistently repeat their own name at the end of a comment post, as though it was paper mail, despite a) that they've obviously grasped the purpose of the "name" field, and b) it should be obvious from watching everyone else that you don't need to do that.


I'll see your Feersum Enjin and raise you Riddley Walker; reading dialog alound to make it intelligible slows the reader down a lot.

The reason I don't use dictation software isn't the errors or my typing speed or the antisocial nature or the fact that you have to turn the music off, or that I giggle when I think of voice recognition denial of service (gag sacked employees on their walk out of the office so they don't run along shouting File Close No). It's that the words only appear in my head as words I can hear in my head if I'm having trouble writing a paragraph; it's a sign I'm stuck. Usually the words are generated somewhere between my brain and fingers and I get to read them as they appear on screen. If I have to get the words onto my tongue, that's a different experience for me - maybe even a different neural path? - and the style of what I say is different. I assume one gets over the self consciousness of hearing oneself spout prose a la Moliere, but I'll be interested to hear how different the phrasing in your head and out of it is by the end of a novel ;)


As an aside, reference to audiobooks. I'm just listening to an RNIB audio book transcription of Ken McCleoud's Nuton's Wake. It's read by an Australian woman, which is fine once you're used to it. (John Lee can't read everything.) But when She's reading the Scottish character, lucinda Carlile's speech in an assumed accent. It'ts quite ircsome. A special difficulty with audio books when voice artists come across inflected accents. How with out attempting to do that accent, would they pronounce the speech. And alas, if it's not convincing it does detract a bit. my internal accented imagination's voice is perfect of course. providing I've heard the accent in question somewhere IRL. ;)

I can't remember her name now but a Scot was the reader the for the audio version of Halting State.


I enjoy variants on the usual dialogue punctuation, the only rule I insist on as a reader is consistency so that I stand a chance of learning the rules. You got your Cormac McCarthy and your Samuel Beckett all dancing their own way. Feel free to fire for effect.

I have a friend that uses Dragon to vomit up large tracts of thought and then goes through a lengthy editing phase. I would think that doing that for just dialogue would be more productive (picture: black leotard and the actor's stance while prancing around with a headset mike). Style changes, well you'd have to think that style change is inevitable - in the words of one of two fellas standing on the edge of a cliff.

"You go first."


Personally, I'd be worried that having a theory of writing would abruptly end whatever minuscule ability I may have at writing. The caterpillar problem. Good luck, Charlie!


I don't know how familiar you are with neorealism and verism from the italian cinematography and literature, but both of them tried a lot to be as much "true" as possible, the first by also using common people as actors and natural speech.

The result, while powerful and of unquestionable artistic value, is not always really easy on the reader/viewer and for sure not to everyone taste.

What I mean, all those conventions and stereotypes do exists for a reason...


Two things:

If you really want to get tortured by ethnic variants of English in fiction, try Henry Roth's Call it Sleep, where he's trying to recreate all the immigrants' dialects in pre-depression New York City from Yiddish to Irish.

Secondly, in my brief career as a reporter I found that recorded speech was totally different from what I heard when interviewing someone. We do a lot of unconscious editing in our heads. I don't mean we necessarily change the meaning of what people say, but rather we eliminate the stops, starts, parentheticals and so on in normal speech. I can remember thinking that someone was very articulate, but when I listened to the tape recording and tried to pull quotes, it was gibberish. On the other hand, there was a guy I interviewed regularly who spoke in complete sentences and paragraphs. What he actually said corresponded exactly with what I remembered he said. I like using him for that purpose, but he sounded unhuman to listen to.

Have fun with your experiment


The experiment sounds interesting, but I suspect you may have to re-learn the art of creating different characters' phrasing, to some extent. But the only way to find out is to try.



If you didn't know already, I can tell you that Australian novelist John Birmingham has written two novels using Dragon Dictation software. He pretty much had to write After America using it after he busted his forearm awhile back.

I don't think he'll mind my mentioning that I saw some of the raw copy in my role as research consultant. While there were problems with transcription errors, I didn't think they were any worse than what you'd find with typed documents. Additionally, I felt that the narrative flowed much more naturally via dictation as opposed to the previous typed documents I have read.

That said, or typed actually since I am not using dictation software at work, I have only Birmingham to reference as an example. It works well enough that I am seriously considering using dictation software for my own efforts in the future.

Good luck with it.

Respects, S. F. Murphy On the Outer Marches


The following quote from Daniel Everett in the 16th March edition of New Scientist (article here if you've registered) may be of interest:

As societies adopt a written language for cultural reasons, their grammars often change. Perhaps, more accurately, they begin to adopt a second grammar - a grammar of written versus spoken language. Many studies show written and spoken language differing in numerous, often profound, ways, such as sentence length, complexity of paragraphs and so on. The new features of the written language are caused by alterations in the expression of syntax, owing, ultimately, to the cultural decision to write the language.

Your comment regarding reporting interviewing people reminds me of what reporters used to do with Canada's former Prime Minister, Jean Chrétien.

His syntax was notoriously mangled in both English and French. When covering a press conference during which questions and answers were given in both languages, reporters often took his responses in one language, translated them to an intelligible sentence in the other language, and reported that as what he said.

The joys of living in a bilingual country.


I gave up reading Banks, after Feersum Enjin. For some reason he is very hard to read (even when not writing in phonetic style). I suspect something to do with me being dyslexic. I can't spell myself, but struggle hugely with non-standard spellings in my literature.

Its an interesting experiment, the other thing about dialogue in books, is that it also removes all of the non verbal ques that humans use (shaking/nodding/inclining of head, position of eyes, eyebrows, shoulders, arms, hands gestures, smiling/frowning etc), and while some of these can be added in description, its (I think) a hard thing to communicate well.


When I was very little, I used to speak all my stories out loud to myself. I just wandered around the house speaking in different voices -- my characters' voices. I'd do the same scenes over and over, until the echo in the room sounded the same as the echo in my head. I did this until I was comfortable typing.

As an adult, I still do this to an extent. I subvocalize some dialogue as I'm going along, but in the thick of a project I usually wake up -- and go to sleep -- hearing and rehearsing conversations between characters. I know what they're going to say in an important conversation months before I write it. This also makes writing new scenes easier, because I have a sense of how these people interact on a really basic level and I don't have to bother with wondering whether or not X or Y would "really" say something.

The trick, though, is that I do this everywhere. Not just my office or bedroom or other writing space. I do this in the shower, in the kitchen, on my yoga mat. So for a program to really hear everything I have to say on the matter, as it were, I'd have to carry the associated device with me to all those locations, or install an apartment-wide surveillance system to bug all my conversations with myself. And that would just get ugly.


David Weber uses word recognition software? I can totally believe that.


I hope you have some acting experience, or a lot of experience in writing radio plays or plays for the stage.

Otherwise, all your characters will end up too much like you.


"we don't think (or communicate) in formal grammatical sentences"

I've noticed that as well --- isn't it terribly interesting that "formal grammar" (aka, the only kind) is a side-effect of literature? In fact, I believe that there exists no written descriptions of grammars before the early modern period. Latin "grammars", for example, weren't a set of production rules, but lists of incorrect forms (in other words, heuristic descriptions of the low-class register -- "One says hello, not 'ello".

And most of linguistics is predicated on the existence of "grammar" as a real, universal cognitive entity. Hmm...


A lot depends on the writer's training. All the Perry Mason books were dictated because Erle Stanley Gardner was an attorney and was used to thinking that way, and I don't think his narrative voice suffered because of that particular quirk, anyway (we may all have opinions on the quality of the Perry Mason books, of course).

Practice may not make perfect, but it often does make better; I think as more writers become used to speech recognition software, and as the software improves, there may not be as much difference as we currently see between typed and dictation writing.

Admittedly, I do have a prejudice here; if the practice eliminates or slows down the page-long Faulknerian sentence, I for one will not mourn.

I hear characters' dialog out loud in my head anyway, along with the full video presentation, which I can generate far more cheaply than say, Hollywood can. I can't wait for the technology that will let me do it.


I really like the technique of including official papers along with the regular narrative. It's a nice way to include necessary exposition without feeling so much like an infodump. There's also that clash between the dry, analytic description of events and the immediacy of the memories of the person who lived through it.

Getting the right ear for dialogue is tough. The original Star Wars is a very quotable movie but it's astounding to see the difference between the shooting script and what made it on film. Most of the lines were twice as long in the script and would be impossible to actually speak properly. As Harrison Ford said: "George, you can type this shit, but you sure as hell can't say it." The very act of running lines in practice paired things down until they were punchy and memorable.

I think the use of observed detail can help fill in the gaps in the dialog.

"No, I don't know anything about it," he said evasively


"N-No, I don't know anything about it," he said, a little too quickly. He started to tap his foot and stopped a second later. The man was sweating the moment he walked in. This wasn't going to take long.

A lot of people hate dialog tags. Stick with said. Don't get cute.


Do people who use speech input a lot end up talking like Severus Snape?


The 'Feersum Enjin' phonetic sub-plot is interesting because it causes much more pain to readers than the differently variant (but still pretty weird compared to straight English) writing in the Barbarian's sub-plot of 'The Bridge'.

I think this is because Bascule's prose is much less of a licensed unorthodoxy (if such a thing can be said to exist) than the Barbarian's - so it's both harder to read and conjures up a less sympathetic archetype in the reader's mind (I still find it --very-- hard not to think of Bascule as some kind of idiot or leet-speaking script kiddy, even though I know he isn't).

'More work' combined with 'less sympathy' means a much higher risk of the eight deadly words being uttered at some point by the reader and I do get a sense that Iain Banks was pushing hard to see how close to the line he could take things with Bascule's sections.

WRT the signing off thing, I do it because: (i) my online handle is not my given name (although it is derived from it) (ii) I'll sometimes post as anon to blogs/LJs/whatever. I'm not hiding - I just can't be arsed to sign in always, but my habit of signing off IDs me to those who might be interested in tracking my thoughts (a population consisting of me, basically - but then I'm a raging egotist) (iii) I'm a refugee from usenet, so not having some kind of tag at the bottom is just --wrong--

So for me it's a mixture of kinda-sorta needing to do it along with a hefty dose of affectation/personal style or whatever.

Regards Luke


I think grammar is probably best thought of as an emergent property of a shared spoken communication system. I suspect it's a touch more probabilistic than would be suggested by the idea of 'formal' written grammar. If you're a glutton for punishment, try

Side thought. There is something particularly odd about speaking in circumstances where you know what you are saying is being transcribed. I've appeared in front of a Parliamentary committee and knowing that everything I was saying was going to be written down had the effect of making me talk more as I would write. Watching myself later, I couldn't help thinking I sounded horribly affected and as if I was reading from a script, though I wasn't, but written down it might have helped.


I find myself wondering how it will turn out from an interested point of view. But I suspect not incredibly well. Otherwise I'd imagine a lot more playwrights, particularly those writing for the radio, to use it - after all dialogue is absolutely king there.

But... no gestures still so you're writing a little artificially. TV and film scripts need to mention too many things about the shot structure I'd have thought.

Although I'm an educator, I never really looked closely at early years stuff. I wonder though - once we are over the hump of the worst of learning to read how well do we take to reading dialogue? Anyone happen to know? Or has verbal storytelling in the past (parents reading stories to their children in bed still happens in some houses after all) already adapted us to a different expectations about how characters speak in a story I wonder?


I'll be interested in hearing how it comes out, as I've listened to the last five Charles Stross novels I've "read" (after reading the first four in the usual way.) The evaluation will be complicated, of course, because there'll be a voice actor and possibly producer's ideas of what it's supposed to sound like in between Charlie and his editor, and me. But I'm going to be interested none-the-less.

Since audiobooks allow me to keep up my normal reading load while dealing with all the additional responsibilities which have come with middle-age, I spend a lot of time wishing better screen readers existed, or that I had time to work on them myself. I've got the software chops, but, well, see above about the copious nature of my free-time.


I've heard a rule of thumb for writing dialect that requires cutting the differences between dialect and common language down to about 10% of the spoken version. I think this also works for more "naturalistic" dialog: use 10% of the em's and er's, and sprinkle in a bit of punctuation to indicate breath control (but you probably want to go light on the ellipses; they get annoying to the reader (to me, anyway)).

I look forward to your report on how the experiment goes. I suspect you'll find that you'll have to cut back a lot of the disfluencies in speech (pauses, interruptions repeated words and phrases, restarted sentences, words with elongated pronunciations, and filled pauses such as uh and um) in the second draft. But that's just my hypothesis; your experiment will tell the tale.


"Or has verbal storytelling in the past (parents reading stories to their children in bed still happens in some houses after all) already adapted us to a different expectations about how characters speak in a story I wonder?"

I think you'd have to look at the structure of stories told in non-literate societies. I don't think we can make that distinction within our own --- we're thoroughly enmeshed in our narrative traditions. TV, radio, how we tell stories --- it's all after 500 years of the written word being dominant.


I've tried something similar. I bet you'll find this as frustrating as I did. What works with perfect comprehensibility said aloud often becomes disturbingly ambiguous rendered as bald print on the page. This isn't a matter at all of umms and ahs and redundancies excised, or the lack of gesticulation or facial expressions: rather intonation, rhythm and pitch do an amazing amount of work in spoken language, and it's my experience that written dialogue needs significant distortion in order to pick up the semantic lag of vocal but extralexical morphology.

For example, you must have had (or observed) all sort of fraught encounters over misinterpreted emails and text messages, which someone or someone's interlocutor intended entirely respectfully, calmly, and not at all as an attack, but which nevertheless came across otherwise. But, somehow, those very same words would have been fine, if they'd been spoken by phone! You'll find, I think, that you as the writer will hear your own phrases as you intended the delivery, but others will read them differently. I guess we'll see, right?!


Feersum endjinn has been the whipping young person here too many times to just ignore the abuse (and I loved the novel, too).

If you can't be bothered to activate your cerebral refactoring endjinn maybe this book really wasn't for you. I think the Bascule's narrative was remarkably consistent, so much so that I'm still wondering if it wasn't written in plain English and the relevant parts later translated to Bascule-speak. Easy to pick up, once you've struggled with the first couple of lines; a simple case of reading it aloud to yourself in your head.

Much easier to get used to than spoken (northern?) English near Manchester, which is English with German-pronounced-vowels to me.


Have you read The Adventures of Huckleberry Finn by mark Twain? The dialogue reads very naturally; I think that was an aim of the author. Anyway, if you were looking for an example..


Since I only see one recommendation of William Gaddis so far in the comments, I have to add a second. He manages voice so well that he dispenses with attributing dialog entirely, and yet you quickly learn to recognize who's speaking by voice, despite unannounced scene changes, etc. Character A gets a phone call from character B (who doesn't identify himself), A hangs up, B keeps talking to C, and you just hear their voices in your head. It's like magic.


I wrote a character who dictated a diary into text, which included complex punctuation (such as clauses and semicolons). The only problem is it seemed too well-structured to be spontaneous dictation. So I wonder how sophisticated a speech recognition program could one day be, more like the real Siri from Blindsight – able to interpret highly complex speech patterns and transcribe them even if there’s no real understanding of what is meant? When it comes to dialogue (something I tend to have trouble with), perhaps it can keep a check on the consistency of a character’s speech idiosyncrasies (“no, he doesn’t tend to use slang in that situation.”), or would that be taking away the art of the writer?

As an aside: I've been criticised for only using single (‘’) quotes in dialogue. My theory is that this is more of a British thing and not just because modern text tends to be larger and clearer. Also it’s much quicker on the keyboard if you’re not an especially proficient typist such as myself. That being the case, typing is still my preferred method since it helps me better to focus my thoughts. Perhaps it’s matter of habit.


On a completely different topic, just been watch a TV series set during WW2. Looks like every vehicle back then was kept in immaculate condition - all highly polished, no dents and no rust.


I remember that when I was reading Feersum Endjinn I had to make a conscious effort in Bascule's parts to read fast through those. At high speed, some sort of mental error correction kicked in and took care of the weirdness of that transcription together with (some of) the errors I introduced by reading too fast. I also remember thinking that it was very systematically done, to the point of wondering whether it could be done as one (large-ish) regular expression, and if so, whether it was an invertible transformation.


On the other hand, there was a guy I interviewed regularly who spoke in complete sentences and paragraphs. ... he sounded unhuman to listen to.

Isn't that what media training is all about? I've spent too much time on the receiving end of hostile media attention and one thing I learned very early was to do exactly that. Also to slur slightly between phrases and use word ordering to make misleading edits harder.

There's a whole style of speech required when the powerless speak to media, and it's both harder and less natural than what politicians do. Politicians have enough power that egregious falsehoods aren't worth it for media, but little local people? "I don't think this is safe, someone's going to get killed" can easily become "hostile activist threatens that 'someone's going to get killed'". Seeing myself say that on the TV news only had to happen once to make me decide to change my approach to the media.

Whole sentences can also be a useful rhetorical technique, as they happen faster than most people can process so they're much harder to argue against. The trick is to keep it just barely faster, or it starts to sound like a prewritten mind dump that should be ignored.

I also remember thinking that it was very systematically done, to the point of wondering whether it could be done as one (large-ish) regular expression, and if so, whether it was an invertible transformation.

That's what I was thinking shortly before I started skimming it rather than trying for any sort of comprehension. Because if it's an invertible transformation, it could be inverted and thus yield a narrative written in something approaching normal prose. Destructive to the author's vision, maybe, but at least it wouldn't resemble something written by a script kiddie or textspeak addict, and it wuld be much easier to read.

I also speculated about obtaining an ebook of Use of Weapons in order to shuffle the chapters into chronological order, to see if they made more sense that way.


"Feersum endjinn" to me was an extension of Mary Daly's "Gyn/Ecology". Daly was explicit about it - her book is designed to be unreadable to men but readable by women. In practice she overdid it, resulting in a book that many women also find unreadable. Apparently both are quite good books, and if english translations are ever published I'm keen to read them.

I think it actually comes down to reading style. If you read by word shapes that style is gibberish, but the more you read by stringing letters together as sounds in your head the easier it is. It's like the Asterix books - all the name puns only work if you sound out the names. But people who read a lot, and fast, tend to read by whole word shapes, which means that they are dramatically slowed by markedly variant spellings. For me, it's more than ten times slower to read the variant spelling, which means the book has to be more then ten times as good to be worth the time spent. And sorry, but Iain M Banks is not 10x as good as Charles Stross. So Feersum Endjinn or Halting State is no contest.

I similarly dislike the chopped mince approach to dialect. Sure, the actual words are in there if you pick at it enough, but why would you bother?


We're actually very good at modifying our speech patterns for our audience, particularly if we've had a bit of practice.

I forget where I heard or read about an experiment ---at MIT, IIRC--- where they were trying to let a computer pick up natural speech patterns by letting it talk with random passers-by. They had set it up as a campus information kiosk, and people could ask it questions about things like class schedules, directions, weather reports, and cafetaria menus. After a few days, long before the computer picked up sophisticated speech patterns, users had found out that if you strung together some words that the computer recognized, without any of the little words and conjugations binding them together into a proper sentence, you'd get an answer that was useable, instead of an interminable conversation. Since the computer was programmed to pick up speech patterns from humans, it quickly started answering in similarly simplified grammar, reinforcing the user's tendency to phrase their questions that way.

Anyway, that anecdote makes me think the proposed experiment will be very interesting, in some ways even more so if it fails than if it succeeds. Maybe there's even an "uncanny valley" where dialing back the voice recognition system, forcing you into a less natural speech pattern, is less disturbing than making the voice recognition the best we know how?


It's like the Asterix books - all the name puns only work if you sound out the names.

Except that it's a lot easier to go through Asterix again and sound out only the character names. The third time, after you've gone through it a second time to see what Idéfix is up to on each page, of course. Unless you're French and old enough to recognize the characters as caricatures of public figures of the time when the album was drawn, in which case the name puns are as likely as not to provide some extra laughs with the caricature (or so I'm told) the first time the character is encountered.


Don't get your hopes up.

When my wife was in graduate school, I transcribed hours upon hours of oral history tapes. Later, as I lawyer, I've had occasion to read a lot of transcripts of oral communications (not always in the tame zoo of a court room).

Transcribed speech is ugly and boring in all but the rarest cases. When you are in the moment experiencing speech, you subconsciously edit out all the crap and process what you think you heard and not what actually came out of the other person's mouth. It is stunning to go back and look at seemingly perfectly understandable conservations only to discover just how incoherent they were.


Well Barbara Cartland wrote most or all of her gazillion books using an early beta of Dragon Dictation known as a "typist", and she did ok :)

I used to dictate quite a bit for work. I found it worked well for things where getting the sense down was key and little fine tuning was needed, but not great for pieces of work needing refinement. I can see how it would work for dialogue though. Interesting experiment.


I tihnk it atcually cemos dwon to rdaeing sytle. If you raed by wrod spahes taht stlye is gibrebish, but the mroe you raed by sirtnging lrttees together as snuods in yuor haed the easeir it is. It's lkie the Astreix books - all the nmae pnus olny wrok if you snuod out the nmaes. But poeple who raed a lot, and fsat, tned to raed by whloe wrod shaeps, whcih mnaes taht tehy are dramctiaally slwoed by markdely varaint siellpngs. For me, it's mroe tahn ten temis swoler to raed the viraant spenlilg, wcihh mnaes the book has to be mroe tehn ten tmies as good to be wroth the tmie spnet. And srory, but Iian M Bknas is not 10x as good as Clarhes Srtoss. So Freesum Ennjidn or Haitlng State is no cotnest.


In speech ( and writing ) often I cannot articulate an event or object. Once I've tripped up I can see the thing I am discussing in my mind's eye, but there are no words associated with them. † Frequent temporary placeholder tags include; 'thingy' for a person, frantic hand waving for a verb and 'whatsit' for a object (also dodah). My mother then says, "Don't use that word 'whatsit' say what you mean!" I reply, "If I knew what the word was I wouldn't have to say whatsit!" After a little bickering the word eventually appears and we can carry on where we left off. I suppose it's being thrown out of the 'fiction' of the dialogue that's irritating. Although this is natural speech it doesn't really drive the narrative along and I can tell you it gets old real fast.

† This includes concrete words like frying pan; I vividly remember pointing at the thing one breakfast time and still not being able to speak it's name. I don't bother often with Proper Nouns because they've flown off never to be seen again. In writing the words I struggle with are usually more subtle and abstract and strangely get tangled up as vague homophones. I also say colours wrong and will deny all knowledge of, for example, saying that pillar boxes are blue, because the word in my head was red. It is unsettling when you can't rely on what comes out of your mouth. It also depends on how tired I am.

In my brain I feel that 'stuff', the words associated with them, their spellings and the sounds they make live in different places that are only tenuously connected. While I was rehearsing this comment as I was reading the previous comments, I've still had to rely on it coming silently out of my fingers to make it work.

It's always a great relief to be told that standard grammar bears little relationship to how we speak or think.


Yes it's slower, but also, for me it forces out the word sounds in my head rather than the meaning ( content ). In the original comment I had a clear picture of leafing through the book looking out for Dogmatix. In your comment I only had a pale shadow of that, probably because the content was familiar to mean already, but it was quite load in my head. I can tell Im struggling with my reading when the words become noticeable.


Not surprising that Carrie Fisher later served as script doctor for other movies.


It will be interesting to see if the result is noticeably different. Will you be posting any samples here for us to see as you go along?

However, I'm sceptical about how different the result will really be for a couple of reasons:

Firstly, correct me if I'm wrong, but the speech-to-text programs are (understandably) far from perfect in their transcription, so - is it really going to pick up the messiness of actual speech in a way that's understandable/readable/translatable back to speech when written out? Seems that the result would be paragraph-length sentences with a ton of commas, semi-colons and dashes in them. Possible, but tricky, to get right when you are able to plan it out on a page, but surely impossible for a speech recognition tool to do well. How can it tell the difference between a pause that means a comma and a pause that means a colon? And following on from that, I presume therefore that the raw transcription actually requires quite a bit of editing to turn it into a reasonably-polished output, and in that process, aren't you right back where you started in the sense of creating artificially neat and pared-down speech by writing it out?

Secondly, as those funny bits of text where the letters and words are jumbled/missing have been made to show, what we see on the page doesn't bear all that much relation to what we hear in our head. We can read very mangled writing completely fluently, with it sounding neat and polished in our internal monologue. Look at Feersum Endjinn - bloody annoying for a chapter or so but after that you get up to speed, and for the rest of the book don't even notice the odd words anymore (thus making it, a bit pointless. I always disliked that idea - effort on both the writer and reader's for no actual result). Same with the phonetic Scottish-accent stuff that some Scottish (-dwelling) SF authors seem to keen on recently - if one is passingly familiar with how, say, Glaswegians do actually sound, you can immediately read it fluently and it's not making any kind of difference in your head compared to how that character's voice would have sounded if you'd just been told at some point that they were from Glasgow (or some SF equivalent) and had that accent.

edit: bugger, seems people have been here before me and jumped straight to the Feersum Endjinn example. Should have read the comments a little more carefully.


I wish I actually could edit comments, incidentally :( Proofreading properly is somehow psychologically impossible until it's been submitted. Like this hurdle you have to get over.


That's what the preview button is for, seeing the text in a slightly different layout makes the proof reading easier.

oh and further to my comment on dirk's comment: the mangled words that look like real words don't translate back into the 'proper' word.


Kevin J. Anderson is constantly going on about how he dikta-hikes, all of his "writing" is done with a recorder as he hikes, "near perfect prose."


Many years ago, before the iron curtain had come down and when English was much less ubiquitous in central Europe than it is now, I met a man in Prague who had taught himself English almost entirely by reading science fiction and detective stories. He was extraordinarily fluent - but he spoke written English not spoken English, and the effect was quite surreal. Dialogue in written language is not a simple transcription of spoken language and the second order translation of written back to spoken brings out just how different they are. I don't think that is something anybody need worry about, or that there is much to be gained by some artificial realism in the rendering of dialogue.

But then I thought the language device in Feersum Enjin worked brilliantly.


And - noticed this last night on a 20's tec series - there are only cars, no horses, which while very prevalent at the time are too expensive to use as extras till you get back to Victorian times.


The basic difference between real speech and art speech isn't formality, it's length. Virtually any serious interaction between people other than armed robbery runs to a wordlength that, even with the timesaving potential of reading, nobody would tolerate. People in novels, films and plays jump directly to their info dumps without ten minutes of smalltalk. Lovers break up,politicians issue revolutionary manifestos, armies march in ten minutes or less, and we don't find it odd despite being unable to see someone to their car after a social dinner in under thirty. Dramatic structuring involves a quite inhuman freedom from distraction.

Which leads to the possibility that Charles' experiment may shift him to even longer books. Reamde footlocker size, here we come.


I had no trouble with Feersum Enjin. I've always wondered if that was at least partially due to my being from the same neck of the woods and that Iain, in transcribing phonetically, actually managed to capture some of the qualities (though not necessarily the syntax or the punctuational use of "fuck") of the accent. Certainly, when I read the book, the mental voice I ascribed to Bascule was the local accent, albeit a little stronger than my own.

The rationale I've heard is that Bascule's stuff is phonetic because he's dyslexic, and it's thus a machine transcription of his speech. Given the issues Siri has with Scottish accents, maybe the problem will never be solved, and in the far distant future, they'll just have given up and decided to transcribe it phonetically. The occasional voicemail my Mum leave's on my Google Voice line gets transcribed so terribly it's not even fun. The most recent read:

"Hi darling, it's just mom let you know if they said they feel you like softy kicking sometime. 9 6 9."

I've never listened to the message, so I've no idea what it was meant to say, although the translation of "mum" (which has a very clear short "u" as in "underwater" sound with the accent) still got transcribed OK.

I find Banks' regular written style (particularly in the more recent SF) very readable, almost chatty, as though the guy's sitting beside you in the pub telling you the story as though it happened to some friend of his. It does jar occasionally, but I like it. Not going to make a comparison with Charlie - both are fairly informal writing styles, but they're distinctly different.


A writer friend of mine, who writes longhand and transcribes using voice, uses a great technique (suggested by a mutual friend) for non-standard names and words: use words that are in the dictionary, but in an unused class (e.g. names of countries), in a one-to-one code. At the end, search-and-replace. You just have to remember the code. :-)


It turns out that a lot of those disfluencies I mentioned in #55 are meta-signals or out-of-band signals that communicate e.g. emphasis or that the next word choice has a long access time. For instance, "um" can indicate the importance of the next phrase, or can denote "please don't interject, I'd like to continue on this subject for a bit."

See, for instance, "Hesitation in speech can. . . um. . . help a listener understand"


I actually did that recording-a-conversation experiment earlier this year for an EFL course, and I completely agree with Charlie. You can see the conversation here

As to the experiment, it sounds really interesting, but won't that mean you have to write all the scene description, and then go back in and dictate the dialogue (or vice versa)? When I write, I usually have to switch rapidly back and forth between the two. Is that technically possible?


Reading aloud lets you feel the flow. Even of a no-fiction..


Likewise, at least (early in "his" second chapter") once I twigged that Bascule was best read phonetically.

As for the suggestion that it was done using regexps, I'm just going to say s/keyboard/leopard .

Personally I prefer to use a full-size desktop leopard, regardless of which platform I'm leoparding text onto!


Which leads to the possibility that Charles' experiment may shift him to even longer books. Reamde footlocker size, here we come.

No risk of that.

I'm held by contract to write no more than a certain length, and although the contract is elastic, there are constraints at the publishing end imposed by the cost of printing and binding a bigger book: simply put, it's less profitable to them so they discourage writers from doing that.

There are also costs to me if I try to write a Stephenson doorstep: namely, even if I know what I'm doing (and long books are structurally different from short ones, in subtle ways that make them harder to write) it takes roughly three times as long to grind out three times as many words, so I need to have three times as big an advance to live on while I write it. If it was guaranteed to be a best-seller then of course $PUBLISHER might very well be willing to fork out three times the cash up front, and accept higher manufacturing costs for the dead tree edition ... but you may over-estimate my popularity.

I will confess to occasionally harbouring a fantasy about being able to re-package and edit the Merchant Princes books into the nearest semblance to their original planned shape that I can come up with -- merge the first two, then do books 3/4 and 5/6 as doorsteps too, so a trilogy of 600-750 page books (minus about 50-100 words of now-obsolete connective material in each volume) -- but that's not going to happen unless I find a new publisher for them, which is about as likely as spotting a formation of airborn pigs turning on final approach into Heathrow.


Ref para #4 - Speaking of which, and noting that you'd need something really spectacular to top the existing climax (see, no spoiler), is there a business model other than finding someone (individual megafan or a publisher) prepared to pay the sort of advance you presently command that might make actually writing MP 7 thro 9 make economic sense for you?


Business models: I am keeping this under review.

Please bear in mind that I have limited time and limited ability to write. I estimate I've got maybe time for another 30 properly written books ahead of me in my working life. That may sound like a lot, but the sad fact is that as one gets older it gets harder to deliver new and interesting stuff -- I could settle down to grind out more-of-the-same formula novels, but that's not what you read my books for, is it?

So. Suppose I've got 30 novel tokens. What should I spend them on, bearing in mind that I want to (a) live comfortably and (b) get in my best shot at literary immortality?

The sad fact is, the Merchant Princes books were brewed up during a tight patch in 2002 when I needed to make money fast. Instead of which, they sort of flatlined. It's very hard to apply the defibrillators to a series that's already six books in, and each MP book I could do would be one less Laundry book ... meanwhile, I'm now less hunger-driven and more interest-driven. It's not certain that, even if you waved an infeasibly large brick of £50 notes under my nose, that'd be enough to tempt me into writing another three of them instead of, say, three Laundry novels that paid less but were more satisfying to write. I'd need to also have a solid idea that needs exploring via more Merchant Princes books, not just an income stream.

(I have a very vague idea for such a theme, but it's still a bit nebulous. And it's a bad idea to volunteer to write a novel or series solely for the money: you can spend a very long time being excruciatingly bored if you go that way.)


"I could settle down to grind out more-of-the-same formula novels, but that's not what you read my books for, is it?"

Works for Jim Butcher and me. I think he's up to about 13 in the series.


Not entirely; that said, not everyone I know is a fan of yours, and the least liked books IME are the hard SF "idea soup" ones rather than, say the Laundry or the MPs. Actual sales may differ from my experience.

Personally, I'd like to see the "World 3" English empire explored further rather than the "World 1" American empire (I have Eric Flint and John RIngo for that ;-) ).


Since I met Charlie and read his books, I have been keeping an eye out for them around the place. The SF and LAundry ones are now widely available across Britain in bookshops, but I have seen fewer of the merchant princes series. My local library has the MP series, and they appear quite popular, and they also have the SFnal books.

Meanwhile, second hand and charity shops have been including more of your books over the last 3 or 4 years. Usually only one or two, perhaps someone bought it and didn't like it or a student needed to clear some space. Either way Stross is percolating through the reading public across the entire country.


I'd need to check to get actual figures for the MP each way, but I know some of my copies of them are Tor North American copies, and all the other Stross volumes (I mean all volumes available, not just the selection I've bought) are UK imprints.

This may explain why the MP are less common in UK general bookshops. I was actually introduced to the joys of import SF books by an SF bookshop.


Sell Out, Charlie! Go wild, pull a clancy and franchise yourself up the wazoo! Ghost writers! Crap Video Games! Dubious tie-in merchandise and patronising young adult books!

If you wanna go the whole hog, pull a Cussler and write yourself into each book, as the mysterious and affable geek who supplies the crucial information to save the day!

Screw interesting ideas or artistic integrity, you could be lying in a gold plated helicopter jacuzzi in the andes, Right Now!


Wouldn't pulling a Cussler mean that the main protagonist in each of Charlie's books would drive a Volvo? ;-)


MP #4-6 have not been published in the UK at all. And books #1-3 didn't sell well and are out of print.


Also, he may have to call any possible children Bob.


"we don't think (or communicate) in formal grammatical sentences"

I do, at least since I learned to read. I've had a few people try to make fun of me for it over the years. I've also talked to people on the phone or in person after months or years of talking to them via email; several have shown surprise that I sound the same either way.

I read much more than I speak or listen. Though some people claim people naturally think in pictures, any "pictures" I deal with are text. If you say "tree", I think "t-r-e-e", not a picture of something with leaves.

Being partially deaf, I'm quite aware that most people don't speak in grammatic sentences. Not only that, many of them use incorrect words, randomly substitute vowels, leave words out, or just slur the whole mess together in an uninflected gabble.

Given the amount of trouble I have doing "speech recognition" in my everyday affairs, I'm more amazed than annoyed at how well speech recognition software works...


It takes a few moments for me to get tuned in, but it's easier than, say, Vermont English, or the hip-hop-gangsta TV dialect spoken in so many inner-city areas.

speech fragments

The Nixon Oval Office tapes are extreme examples of that; there might be four or five people having a meeting, but it sounds like you're only hearing one side of more than one telephone conversation.

dictating stories

What you hear on playback might be interesting. Asimov was astonished to hear himself shouting at the microphone when he thought he was merely emphasizing a passage.


Do the rights revert to you if out of print? If so you might squeeze some cash out of them by making them a cheap eBook download.


I do mostly speak like I type. Of course I don't swear much, and don't really use much Doric or Scots as a rule.

Also I don't do much "visualisation in pictures" unless I'm trying to explain something to someone, like the difference between "small" and "far away". It's amazing how few people seem to understand that being at 3x the range means that you need to move 3x as far to induce a certain angluar movement at the origin!


Might be an amusing experiement to dictate to a voice recognition/transcription system and play the result of through a Kindle (or whatever) back into the voice recognition system.

Repeat n times and see what you end up with.


Cheers; I'm fairly certain my MP1 is Tor, but not sure either way about 2 and 3. However, that just seems to prove the argument about UK sales at least.

Anyway, as I say my main interest for a possible MP7..9 is in World 3, which could be written and sold as "an alternate history where the American Revolution failed/didn't happen in 1776" (dependant on marketeers).


... or Mo.


Yes, but that can take years. More to the point, the UK rights were sold by the US publisher who'd bought world rights. Trying revert the UK rights would mean yelling at the US publisher and getting their sub-rights department to separately revert the UK right from the UK publisher.

Ever tried eating spaghetti in olive oil with chop sticks? Let alone doing so while using a different set of chop sticks to hold the ones you're picking up the food with?


Thanks for the heads-up re: Feersum Enjin. Haven't read it, and probably won't unless it's an incredibly compelling story that can only be told that way. Recently finished REAMDE - very easy-to-read page-turner and took no time at all to finish the 1,000+ pages.

Generally not a fan of 'modern literary fiction' even though I have slogged through some (not all) favorite authors' stylistic experiments. I read for the story -- ideas, characters, plot, etc. Authors should decide whether they're 'auteurs' or storytellers.

Noticed that some recent Hugo nominees/winners seem to have rediscovered or are playing with over-the-top 70s existential/angst-y/New Wave/woo-woo style. These authors also tend to be in the younger cohort, so maybe these style choices reflect their inner artistic turmoil, angst, etc. (Okay - none of the younger authors that I've read come close to Delany's Dahlgren which used a then-unique text formatting approach - and it worked because the text really did need to be written/shown that way for the story.)

So how does this relate to how different aspects of the book are written - plot narrative vs. dialogue ... When a book relies too much on how it sounds when read aloud, readers who rely on visual descriptors (because that's how they process information) don't have a good story experience. Sometimes, these readers even lose the sense of parts of the story. Same with motion/action. A really good author takes into account the variety of different (sensorial) styles that different reading audience segments have to use to experience a story.


The multi-stranded narrative in Feersum Endjinn works; I'm less convinced that the idea of a dyslexic protagonist relating one of the threads is actually necessary though.


Actually, Feersum Enjinn is a jolly good work of far-future gosh-wow spacetacular fiction, and the price of entry is worth it just for the final sentence.


It depends on the writer's training: Oh, yes. Many literary works before the 17th century were composed -in the author's heads-, then dictated or written in shorthand when they were ready for someone else to comment on. (Acquinas could dictate several works at once this way, and seems to have stopped bothering to write a draft by hand towards the middle of his career). When paper was expensive, pens were awkward, and scripts were slow, the “compose as you write then write a new copy for each draft” approach to editing didn’t make much sense. When text is infinitely malleable, learning to compose a book in your head doesn’t make much sense.


Tricky contracts. One topic that would be of interest to me, and maybe a few others, is the tax position of a professional writer. For example, what are allowable expenses, does IR35 apply, what about NI etc. Also, what about an unsuccessful professional writer (of interest to even more people!). Can you collect unemployment if you are still picking up a cheque for (say) £200 every quarter? And so on...



That'll be why I've had no luck getting hold of them in bookshops then... Ah well, Abe Books here I come...


Yep, MP #1 is reasonably available in London, as are #4-6. #3 is around in a few places, #2 simply doesn't exist.

Drove me insane when I saw all the marketing in Forbidden Planet for #6 when it came out, and literally couldn't get hold of a legal copy of #2 anywhere, even Amazon only had a handful of no international shipping second hand copies.
Tor looks to have finally released ebook versions of them a year or so later.

That being said, I did throw #6 quite hard against a wall when I finished, after buying the others and downloading a copy of #2.
I got a definite impression that OGH had written himself into a corner in that particular world and decided to do a rocks fall everyone dies finish. Expansion now would be ... interesting. I really did like the early part of the tale, if not always the direction it went in.


Part of the problem with phonetic representation of dialects: The writer may not hear the other dialect accurately. Minor example: in Minnesota (and, I believe, Manitoba) "Don" and "Dawn" are pronounced the same. And I'm among the minority of Americans who pronounce "horse" and "hoarse" differently.

Also, people often don't know how their dialect sounds to others.


Depending on which linguistic classification you consult, Scottish is 1) a dialect of English; 2) a separate language; 3) several separate languages.


(1) is just plain wrong. The grammar and some of the individual words are different to English.

(3) is arguable, depending on whether you classify Lallans and Doric as dialect variations of Scots or as separate but related languages. It is more supportable if you classify Scots and Irish Gaelic as 2 separate languages rather than as the 2 dialects of Gaelic.


Frisian: besides the dialect spoken in the Netherlands, there's one spoken in parts of Schleswig-Holstein.

It's the language most closely related to English, though there's been a fair amount of divergence.


A careful reading of IR35 suggests that I'm 100% self-employed -- I meet none of HMRC's criteria for being an employee. The key point is that I decide what projects to offer my publishers, negotiate a contract, then deliver something that is the output of a process I manage myself (using my own tools, on my own time, on my own premises).

Novelists don't get unemployment benefit, but supplementary benefit or whatever they named it to (for folks on low income) is another matter, if their income is indeed below the threshold.

In contrast: a technical author working in an office from 9-5, documenting a company's products on a work-for-hire basis with no ownership of the copyright and a monthly pay cheque is indeed an employee, whichever way you cut the cake, even if there's a shell company in the middle.


"...decide what projects to offer my publishers"

Does that have to be plural? If you only have one publisher would that count as your employer under IR35?


I do have tentative plans for how I'd approach a "Merchant Princes: The Next Generation" series. Clue: it starts up at least a decade after the first series ends, it's pitched as near future SF/technothriller, and a chunk of it in New Britain is set around the time that time line is getting nuclear reactors, supersonic jet fighters, and the very first integrated circuits (i.e. 1960s tech). The gun on the mantelpiece that that series would spring from was placed there with loving care in chapters 1-3 of book #1 ...


Actually, from the single example quoted, I wondered if Frisian might be related to Scots rather than English. "Bairn" is most assuredly a Scots, Doric and possibly Geordie word. If it appears in an English dictionary it tends to be described as a Scots colloquialism.


Nope. And in any case, authors virtually never have just one publisher. (If nothing else, compare my UK, US, and Japanese publishers ...)


Ah, yes.

My wife studied English and Comparative Linguistics in Minneapolis, and in one particular course on US dialects, the class was treated to a certain sentence spoken by speakers from different places around the US.

There was one word which she could not work out. Her spelling of the sound would be 'shown', but that made no sense in the context.

Eventually, she had to ask for the word to be spelt out. S-H-O-N-E.

"Oh, shone!" she said, rhyming it with 'on' rather than 'own'.

At which point, she was asked to repeat the whole sentence herself in her educated London voice, as an authentic example of an exotic accent.

I wonder what the rest of the class would have thought about her original York accent.

(The preceding assumes 'on' and 'own' to be base sounds that you'll all get the same way I do, but I'm not trying to master IPA right now.)


Yes, that last sentence made the hair on the back of my neck stand up (and I mean that literally). As that was the first of Banks' books I read, it set me up to like all of his work, even when there were parts that bugged me a lot (the torture seen in Consider Phlebas and the deeply depressing effect of Against A Dark Background). There's a lot to like in Feersum Enjin and Bascule's strange orthography is a small price to pay for it, IMO.



Preview should involve a large robotic hand that pushes your nose into the screen so you have to read every single word and examine it for correctness.


Would be interesting to put 'Finnegan's Wake' through VRS


The difference between speech and writing is one of the reasons voicemail is so horrible. No speech cues from a third party; not writing.

So you get ", it's, errr, me and it's 4 o'clock and...thingy and I was wondering if you could BEEEEP".


I listen to various podcasts including clarkesworld. One of the things that really hits me is the (to me) mispronunciation of words. Shone/shown is one, but another was with a character name Hereward. The speaker pronounced it "here-ward" but it is actually an old English name pronounced "heh-reh-wood". A bit like when I was listening to Radio One a few years ago late night driving and the DJ mentioned the Greek mythology character Persephone - pronounces as "Percy-fone"


Charlie @ 98 You know why this is: It was the split into #1 & #2 that you told us (long-ago) was at the isistence of the publishers (IIRC) ?

TRX @ 100 (& previous) we don't think (or communicate) in formal grammatical sentences I try to. I certainly try to when speaking, as Charlie will testify (even when I've had several beers) It's one of the permanent after-effects of being not just a teacher, but a science teacher - clear communication is 150% essential.


"we don't think (or communicate) in formal grammatical sentences I try to. I certainly try to when speaking, as Charlie will testify (even when I've had several beers)"

Another piece of evidence --- one has to "try to" speak and think grammatically. If grammar were actually part of the human brain, and not a literate affectation, then those sentences would simply not make any sense at all.

So then, what do we speak and think in? And "fragments" isn't answering the question at all -- a grammatical fragment isn't grammatical at all, given it's degeneracy.


Greg, that was the US publisher. Binding constraints, as I said. The UK market is under different cost pressures; if the series had initially been sold to a UK publish things might have been different.


Okay, I'll give it a go if I can find it at the local book store ...

Out of curiosity ... How does a strong dialect, a heavily idiomatic local English, or dyslexic writing in the case of Feersum Endjinn impact translation (localizations) into other languages? How does someone deaf read a book where much of the text is at best a phonetic approximation of English? (I'm not griping or in a snit, just curious how editors/publishers handle this.)


I've tried speech-to-text and found it difficult to adjust, though in the interest of furthering science, I propose that you write portions of your mixed typed/spoken novel under the influence of an MRI or PET scan, allowing us to see how writing with your fingers or your lips lights up different portions of your brain.

(Think of the marketing opportunities; you'd release the first written work where the reader doesn't have to guess at the author's state of mind. You'll get all sorts of press, and you won't have to commit plagiarism to get it).

Odd Aside: I was going to suggest that screenwriters should find speech-to-text more palatable than novelists given how much heavy lifting dialog does in your average TV show or movie. Then I realized good TV/movie dialog has little in common with everyday speech (Aaron Sorkin's dialog runs about 2x as fast as conversations in real life), so, uhh, never mind.


Speech and reading -

I guess you're aware of the 1922 poem "The Chaos" by Gerard Nolst Trenite cataloguing the vagaries of English spelling and pronunciation. There's a good number of mangled versions on the U tubes where different English dialects strut their stuff and people try to make the best of a bad job.


I like the way Arthur Conan Doyle goes about hinting at accents without getting annoying by overdoing authenticity in the Challenger books. (Many other aspects about them are positively offensive BTW, so don't read them if spiritualism or racism are very distressing to you.)


Ugh, just tried Feersum Endjinn again. Three times now for three tries over about 10 years, I get a mild migraine from the phonetic sections. It's just setting something spectacularly wrong off in my head.

I believe everyone that "it's worth it", but for me, it's just not...


Apparently, when a certain pair of Royal Navy warship firstwere in the same waters, a signal was sent: "At last Antellypee meets Pennyloap"


"Actually, from the single example quoted, I wondered if Frisian might be related to Scots rather than English."

Related to both. (And to Pitcairnese, Tok Pisin, etc.)

Here's what the Online Etymology Dictionary says about "bairn:"

bairn "child" (of any age), O.E. bearn "child, son, descendant," probably related to beran ("bear (v.), carry, give birth;" see bear (v.)). Originally not chiefly Scottish, but felt as such from c.1700. M.E. had bairn-team "brood of children."


Well, yes, spacetacular if you say so. But:

Is there any science fiction in Feersum Enjin, other than the background?

I've been "had" four times by Banks, diligently reading his first four "science fiction" books as they came out between 1987 and 1993. I felt an incredible sense of intellectual "wrongness" as I finished each novel. Something was missing, something was deeply, completely deficient with each novel. But what? He did his characters right, his plotting right, etc. etc. It took me years to figure out that he'd been using science fiction only as a backdrop. He did the literary equivalent of Star Wars but with words, descriptions instead of visuals.

It took me a long time to realize this because he was fooling me with his excellence as a writer. Normally, SF writers who are bad at really using SF themes are also bad writers. They make cardboard heroes and villains, write rotten plots and fail in just about every aspect. In contrast Banks was a success at creating believable, good heroes and loathsome villains, making great dialogue, writing up perfect descriptions, etc. etc. He just didn't understand or wasn't interested in SF ideas.

So, did he change with Feersum Enjin?

Did he finally "get it"?

(P.S.: I find it odd that no one has yet mentioned Kipling's "Soldiers Three", given the amount of made up dialect in it)


Answering that beyond saying that Banksie writes "space opera with characters" rather than "hard SF" threatens to run into spoilers I'm afraid.


I remember when word processors first arrived on the scene, people said novels began to sound like they were written in "telegramese". Personally, I love telegramese. It's a vast improvement.

PS: You're always interesting.


Om language relationships, see

Tangent: Some linguistic theories make political theory seem eminently sane by comparison.


Same with Stephenson's REAMDE - good story, not particularly SF.


I'm curious: what makes you think REAMDE is meant to be SF?

(I read it as a Stephen Bury novel, except $PUBLISHER demanded to use the Neal Stephenson brand name on the cover because NS out-sells SB by quite a margin.)


Alain writes:

He just didn't understand or wasn't interested in SF ideas.

Or you have a remarkably narrow interpretation of SF.

Q: How much "SF" was there in the Foundation series; Dune; A Fire Upon the Deep?


Q: How much "SF" was there in the Foundation series; Dune;

Tons of it, despite statements by the authors themselves, swearing mightily that they based the novels on the fall of the Roman empire and the lives of Renaissance princes. Thing is, you can't transpose the Foundation trilogy or the original Dune novel to a cowboy adventure setting or to a swashbuckling seven seas story. They're too tightly woven with their technologies, and their scientific or anti-scientific mindsets. In contrast I eventually discovered that I could lift the plots from the four Banks SF novels I read or from Star Wars and place them directly in just about any other kind of fiction once I traded the blasters for six shooters, etc. etc.

Q: How much "SF" was there in A Fire Upon the Deep?

A: Can't say for the whole of it. I bought it, started to read it, got bored silly, and then stopped after 20 or 30 pages. In contrast I found at least two other Vinge novels to be quite entertaining and full of SF. Can't remember their names though.


Warning: Off Topic

Vernor Vinge had some nice things to say about Charlie in a recent interview.

Vinge: Probably the most courageous walkthrough into the Singularity was Accelerando by Charles Stross. He actually follows the development from, I think, from the 2010s through the 2070s. He also said that by the time they got to the 2070s, he's no longer seriously claiming that what he's describing would be like the post-Singular world. I suspect that comment was related to the notion that after several decades of this, things would be seriously beyond what a writer could understand in our era, and what the readers of our era would understand.


I hope you're still reading comments, CS, because keeping track of who's speaking isn't nearly as easy as you suggest in the paragraph following "Irvine Welsh". I imagine preceding lines with a name as in a script would be embarrassing as well as clunky, but it's frustrating to have to re-read a passage after realizing I've reversed the participants of a dialog. And while we're on frustration, I loved Fuller Memorandum, but that was a dirty trick to raise the split of Bob & Mo at the beginning without following through at the end.


Back when Algis Budrys was doing his writing columns for Locus (unreprinted, alas) one explained the techniques and experiments he'd used in Rogue Moon. He mentioned how a relative had read the finished novel and, based on the dialog, asked whatever had made him decide to write a book where every character was a gravely degenerated psychotic.


"How much SF is there in [ insert name of author / story HERE ] ??

Everyone really needs to read a superb essay by Ursula K. le Guin in her book "The Language of the Night" entitled: From Elfland to Poughkeepsie

Which really sorts the issues out.

IIRC, somewhere else, she says ... "go ahead, write your story, it'll only be one of the Old Ones, anyway" The much-lamented Charles Sheffield once said much the same thing, about one of his own stories, once he realised he was, in an entirely believable SF setting, re-telling the rescue (or not) of Eurydice


This is off Topic I know but..I have a Good excuse!

Charlie ?? Has something gone wrong with the publishing schedule of ' The Apocalypse Codex ' ?

I've just received a cancellation of my, long-standing, order for the hardback from Amazon ..



We regret to inform you that we have been unable to obtain the following item:

Charles Stross "The Apocalypse Codex (Laundry Files Novel)" .......

Our supplier has informed us that this item is no longer available. This item has now been cancelled from your order #203-1279987-6134710 and we can confirm that you have not been charged for it.

We apologise for the length of time it has taken us to reach this conclusion. Until recently, we had still hoped to obtain this item for you.

You still have some options available to purchase the item(s), or similar item(s): "

EH WOT THe F*K and all that sort of thing!


A P.S. on my previous post ..The Hard Back is still listed on Amazon .com the US of A vians site.

" 1. Product Details The Apocalypse Codex (A Laundry Files Novel) by Charles Stross (Hardcover - Jul 3, 2012) Buy new: $25.95 $16.77

Available for Pre-order. This item will be released on July 3, 2012. Eligible for FREE Super Saver Shipping and 1 more promotion Books: See all items Sponsored Links (What's this?) 1.

All New Persil opens new browser window Find Out About Our Quick Wash Outstanding Cleaning Results 2.

The Laundry Co opens new browser window Laundry, Drycleaning, Ironing, Shirts, Collection and Delivery Advertise on Amazon See a problem with these advertisements? Let us know "

Humm inference and association it would seem that your novel might produce ' Outstanding Cleaning Results '



Cleans up small or large infestations of OLD ONES (TM) ??


I loved Fuller Memorandum, but that was a dirty trick to raise the split of Bob and Mo at the beginning without following through at the end.

I have a term of art for you: Chekhov's gun.

I can assure you that the trigger of that particular gun is not pulled in "The Apocalypse Codex", but I have plans for it in the next book (#5).


That's typical Amazon behaviour.

What is happening is: the British publisher is producing a nice trade paperback, and the US publisher is producing a nice hardback. Both books will be published on the same day.

Because publisher (a) bought US/Canadian rights, and publisher (b) bought UK/Commonwealth rights [minus Canada, of course], they have an agreement not to sell their edition into the other publisher's territory.

However, Amazon's database is a giant sucking vacuum of misinformation. It indiscriminately hoovers up forthcoming titles from all and sundry and throws them up in front of the buying public in hope that somebody will, er, buy.

What has happened is that you have ordered the US hardcover edition via, effectively as an import. The UK publisher then got around to writing their usual email to threatening them with boils and a rain of frogs if they import US editions into the UK, and promptly cancelled all advance orders for the US edition, rather than doing the sensible thing and substituting the UK edition.

What you need to do:

Go to and place an order for the trade paperback.

(If you really want an imported hardcover, you can still get one: you'll need to log in to (the US site) using your Amazon account credentials, and order one with international shipping. As a consumer, the first purchase doctrine means you're entirely within your rights to do so. It's just that, as a wholesaler, isn't allowed to sell you the import directly if there's a local edition.)


this explains, rather neatly, why I'm happy to live in Germany in the case of books (as will happily sell (and is allowed to sell) any version of books in the English language).

For other media - not so much.


On the subject of Scottish language

Scots Wikipedia

Looks like great fun.


404 error. :-(


Fixed it for you :-)




My take on Rogue Moon: There's one character who might have been sane.


Oh, and I thought the people in "Rogue Moon" were perfectly normal for SF characters.


I wouldn't trust too much, judging from what they've done with the Scandinavian languages in the classifications.

By the way, Norwegian for "bairn" is "barn". Scots always looks like a mix of English and badly spelt Norwegian to me.


@ 163 Certainly, 40 years ago, someone from "banks o' coaly Tyne" and someone from Bergen would have no difficulty understanding each other. You also need to look up: 1: "Earldom of Orkney" & 2: "Battle of Largs"


I'm surprised no one has mentioned George V. Higgins, the master of dialogue


Wow. So garbled and yet so easy to read. What really slowed me down was the Iian -- close enough to correct that it looked like a typo. The mind is an amazing thing.


Speech recognition? Unlikely to work )

As a non-native English speaker, I hate it when different dialects are indicated by horribly mangling grammar and spelling.

Well, I suggest looking at it as a challenge. I'm non-native English speaker meself, BTW. Having to read stuff like David Feintuch's "Voices of Hope", with English mangled beyond recognition when coming from tribal lowlifes inhabiting the bottom levels of future New York long abandoned by better-off population, is exhilarating, particularly when you finally get the meaning - and, particularly, the etymology of a word you could not decipher for several pages.



About this Entry

This page contains a single entry by Charlie Stross published on March 18, 2012 1:27 PM.

Gadget patrol: the new iPad was the previous entry in this blog.

Pirate LOSS? An alternative ... is the next entry in this blog.

Find recent content on the main index or look in the archives to find all content.

Search this blog