Assessing forum posts

Being a teacher, I am aware that at the end of the year I'm going to have to give each of my students a grade, and that the more data I have to give them that grade the more reliable and the fairer it will be. It will probably make it easier when I have to calculate the grades too, although a lot of work may have gone into making it easier. Anyway, I like to try to record some kind of score every time I look at anything my students have done. With some activities this is easy. A test, for example, can very easily be checked through, students can get points for correct answers and no points for incorrect ones, in the time-honoured fashion. This is all very well and all very predictable, but most of the real word is not a test with a correct answer.

One particularly successful activity is the forum. Students post their own messages, and then look at other people's and reply to them. It's largely a hands-off activity for teachers as the point of it is for students to interact with each other. It still seems a good idea to assess what they are doing. Ideally the computer would just count up the number of posts they post and the number of words they write, but it doesn't seem to do that, and even if it did, a couple of things would be possible, and less than desirable. One is machine translation. So far, out of around 90 students, only one has used this. It's fairly obvious, as you can see (I've changed his name and place of birth):

My name is Kikai Daisuki.
I am a Honyaku Prefecture Sofuto City coming from.
In the Sofuto high school in my alma mater, a cultural festival has been taken up in the television of NHK.
My hobby is to do sports. I had been doing since especially that the fourth grader of volleyball in the elementary school. It is assumed that it regrets with the first . It will be in the volleyball club also at the university.
I am not good at the study whole.
All [nanode], please teach.
Especially, it is not good.

My best regards in the future for one year.

Some of these features could just be excused as diabolically bad English, for example use of wrong tenses, use of "it" for "I" and nonsensical syntax, although these strongly suggest MT. What made me certain was [nanode]. When computers have no idea what a word means, they put it in square brackets and leave it for whoever is using the system to work out.

The other thing to watch out for is students repeating the same message again and again in each post, although this may be OK if they are typing out the same message again. If they're cutting and pasting, it's probably not going to help them very much. I don't know how much I can complain as it's exactly what I do when I have to comment on thirty students' papers, which share many of the same flaws.

So, as I'm reading all these messages, gleaning some personal information about my students, and also getting some overall impression of common mistakes, I want to give them some kind of score. This could be done by marking positive aspects or removing marks for negative aspects. The simplest thing is to look at accuracy and remove a point if there are any mistakes. Of course these mistakes could be tyopgraphic ones, which may not reflect poor English by hasty typing. It could be a mistake or it could be an error. Many students miss spaces after punctuation, which is clearly wrong, but likely to be something they have never learnt. Spelling mistakes may be typing mistakes. Many students use "let's" wrongly, not realising that the "'s" means "us" so the construction "lets ... with me" is incorrect. Deeper levels of mistakes are in idiom, collocation or usage.

The number or probability of all of these mistakes is likely to increase the more students write and the more elaborate their writing is. Obviously they should get credit for writing more, but on the other hand they should be writing at a level that they can write correctly.