This week I’ve been transcribing the conversations and interviews I recorded while in the Isle of Man. So far I’ve done about an hour’s worth of transcription, which comes out as just over ten thousand words. I have another four hours or so of recordings, and hope to finish transcribing them this week. Then I’ll start writing up my findings. I probably have too much information, but that’s better than too little.
One thing that’s struck me is how disjointed conversations can appear when you write them down. The are loads of repetitions, utterances are often loosely connected with conjunctions like and, so, but or because, they go off on tangents, and often tail off without … Many utterances only make sense in context, and lots of bits can be omitted if the people involved share some common knowledge. There’s also no shortage of interruptions and interjections, and people often finish off one another’s sentences, especially if they know each other well. When you’re speaking you don’t necessarily notice this as much, unless you’re particularly looking out for it.
While it is possible to talk in coherent, well-formed sentences without notes or a script, that doesn’t seem to be how most people talk.
I see what you mean. In the videos section of learnmanx.com, they have some dialogs between Adrian Cain and another guy. These have pdf transcripts which you have probably seen. But I imagine there is a big difference between trying to transcribe what two people say and what a group of people say, especially when they start talking over each other etc.
I can see thats going to be difficult.
Good luck with it!
I served as a Russian linguist in the U.S. Navy for a few years and a large part of my job involved transcribing conversations. Based on my experience, I would say that well-formed sentences were the exception rather than the rule during conversation. This, of course, applies not only to Russian speakers, but speakers of other languages as well. As a result of my time spent doing this kind of work I also became very conscious of how much we rely on various filler words when we speak. When we hesitate and need a moment to think we tend to employ words that essentially serve no purpose other than to avoid silence.
Definitely, transcribing can reveal how disjointed regular speech can be.
I have tried transcribing conversations, interviews, news reports, etc as well, but I have yet to figure out a method to make it a practical learning exercise.
One idea is using a tool like multi-language resource like Euronews (http://www.euronews.net/) where they have certain news segments in video and the transcript below. So you can watch a video in English, French, Spanish, Italian, Portuguese, German, Russian, or Arabic. You can try to transcribe the news for practice and then check the actual transcript afterwards.
Does anyone else do anything like this for practice? Or is transcription more to reveal something about the way people speak?
I also think that’s one of the big problems for language learners. Phrasebooks and dictionaries, even language courses, usually don’t include “filler words”, but they’re such a major part of conversation, it’s difficult to express yourself normally without them. Also to understand other people using them, and know which parts you can filter out.
Yeah, and vice versa, the dialogues in books often don’t look like something anyone would actually say… Many authors just fail to differentiate particular registers.
At the moment I am in the middle of transcribing some data myself for my thesis – conversations by bilingual Estonian-English children and their parents. It’s very interesting but as you say, alot of hard work and does show how differently we write compared to how we speak.
Btw, what transcription system are you using Simon? I haven’t decided on mine yet. It might be CHAT/CHILDES which is designed for child language.
Colm – I’m not using any particular system – just transcribing what I hear as best I can. I’m using the interviews as a source of information rather than a subject of research.
What’s your thesis about?
One tip I got from a veteran fieldworker who had to do tons of transcription:
Don’t work for any longer than 15 minutes at a time. 15 minutes transcribing and 15 minutes in some non-language related activity is ideal. My own experience completely backs that up.
The thesis is an analysis of the speech and language environment of two 3-4 year old children who have been exposed to Estonian and English from birth. The idea is to describe their morphosyntax, phonology, and their parents’ discourse strategy, comparing it to data already present in the literature.
The data is drawn from 80 minutes of recording of the children in interaction with their parents and a background language survey that investigates the children’s language history and exposure.
Try a look at the conversation analysis literature – it turns out that things like talking in overlap, hesitations, etc are very often quite systematic and not simply signs of speech errors to be abstracted away as some linguists or speech technologists might have you believe. John Local, for example, has done some work on finishing off each other’s utterances.
Yes, listening to speech as people actually use it does make one wonder what we mean when we talk about “well-formed sentences”.
There’s one thing I find particularly strange, whereby some people when listening to a story will finish off the last word of the sentence in time with the speaker. This is a personal quirk I think because there are only a couple of people who do it. Is there a name for this one?
Wow, your work is so interesting! I remember sort of doing the opposite and dissecting/translating/analyzing Cree once (armed with a Cree-English dictionary) – it was tough. I can’t imagine how much MORE difficult it would have been if it was a recording of more than one person speaking.