O n f e r e n c e

Results 
Overall what is prominent is that only a relatively small part of the feedback given 
actually fits the sort of classification of feedback often found in studies in the literature. 
As we saw, those classifications are predominantly corrective, typically dealing with how 
language errors are handled. However, the incidence of corrective feedback of that type 
in our study is quite low. The relevant categories in our system are the language ones, 
and the specific item category, since most responses on specific items were in fact also 
negative evaluative and language related and provided what would be traditionally called 
direct correction, e.g. on page 1, "She name Yi Jing Wang" should be "Her name is Yi 
Jing Wang.". Looking overall at table 1 we see that such specific item responses were 
quite rare. Responses to a particular aspect of language, especially grammar, were a 
little more common, but many were nonspecific, e.g. Grammatical errors spotted here 
and there, and even positive, e.g. In general it is fluent. References to language in 
general were more common still at around 15 per 1000 words, but this figure includes 
even less specific feedback like I think you shouldcheck the sentences after you write it 
done. We cannot regard this as due to there being few language features in the stories 
that needed attention, nor to the respondents being unable to spot them, though of 
course there is the chance that some felt their own proficiency was not sufficient for 
them to risk giving specific item corrections.
Much more likely, however, is that they simply preferred to give feedback on other 
things. Having said that, it is not the case that there is little evaluative feedback. In fact 
there are over 50 instances per 1000 words of positive or negative evaluation. Since 
sentences in the feedback are on average around 10 words long, that means that roughly 
every other sentence is evaluative. However, much of the evaluation targets content and 
other aspects like genre rather than the traditional areas of language, e.g. This story is 
quite interesting or I am also confused about the topic. What is also noticeable is the 
relatively high incidence of communicative feedback, in fact half as frequent as 
evaluation. This is often prompted by the message in the content, and at a much deeper 
level, e.g. recording the respondent’s individual response I will hug my mom this week 
because of this warm story or in the form of a generalisation e.g. We can’t choose our 
family but we can choose how to live a life.