Wednesday, December 06, 2006
Transcribing for Free Census
The problems encountered by transcribers can be divided roughly into two groups. Those concerned with the mechanics of FC transcribing; layout etc. And those concerned with making sense of what is written down by the enumerators.
One of the easiest ways of dealing with the mechanics of the task, is to sit back at the end of each page and each ED and look at what you have done. Most of the fields have defaults. In some cases, fields can be blank as a default, but many must have something in them. The county of birth column for instance, Col T, must have a 3-digit code from the Chapman code entered. You cannot leave it blank. Well, you can, but it will then surface as a mistake when I reformat it for checking. If you take a long slow look at each page, you should be able to see where things are missing.
My own personal preference is to transcribe vertically. I do the personal names first, then the surnames, then work to the right, always going down the page. Finally I fill in those details required to the left of the surname. I flag things up as I go if I am uncertain; I also change the text colour to red. If I leave notes, I try to write the sort of thing I write when validating. I have mixed feelings about autofill; but I am a touch typist which makes it easy to type, rather than clicking on the autofill selection. You have to watch the autofill, checkers have complained that people are selecting it without checking it is what they really want.
There is little to be done about the enumerator's writing except hope that you will get better at reading it; or you'll reach the end of the ED and the next one will be better. You can try and improve the image by manipulating it with IrfanView or Photoshop. However, you can also edit what you have done and the fact that personal and place names are often repeated might give you a chance to take a second look at what you have done.
One thing everyone should do is ask themselves - is what I have transcribed sensible? Recently I had a piece where the address was transcribed as "1uoodfbock". Not once, but three times. The transcriber flagged it up. In fact, it was "Woodstock". How many words are there that end in "ood"? Not many.
There are also things you can use to help you sort problems out. The first is Google. Both place & personal names may be detected by Google. Then there are the census returns on line. In Cornwall, most parishes now have 4 sets of census returns online and Wiltshire will slowly build up to at least three sets per parish. There are online gazetteers and sites that list old occupations. However, if you can read what is written, then don't waste time, just type it in. I don't mind you leaving me a note and in some cases it might make it onto the database.
I can see from the notes that many people are worried by the relationships given. Sons that look as though they should be grandsons. However, that is not our problem. Our task is to get down what is written, not correct it. By the way, "son in law" had a rather different meaning in the 19th century. It might have been more correct to say "son by law", and we would say stepson.
One trap that is very easy to fall in to, is to correct the enumerator's words. We are committed to reproducing the exact words the enumerator used. If he is wrong - tough! We have a golden rule "AS IS". If you stick to that, you can't be wrong.
Remember that this is a system. Transcriber, checker, validator and post validator. Transcribers should try and combine speed with accuracy. Don't spend hours over one word; give it your best shot and move on, flagging it up if you are unhappy.