processing of linguistic data while qualitative approach makes possible the analysis of data
with whatever complexity or delicacy. In this study, quantitative approach was employed to
deal with the numbers of lexical errors occurred in the materials while qualitative approach
was used to describe the types and causes of these errors.
 
2.4 Research Procedure      
In May 2012, the subjects were given thirty minutes to write one compositions in the
classroom. To ensure the reliability of the result of the experiment, they were required to
write in the length (120-150) and deal with the same topic. The subjects were encouraged to
write down their ideas and not to spend too much on considering specific words and
grammar. Meanwhile, all the subjects were assured that the result of their compositions
would count against their course marks. The reasons for this kind of design are: 1) to force
the subject to attempt to produce a desired language structure; 2) the subjects have a serious
attitude and high motivation to finish their work. The results would turn out to be better
than being told as an exercise or experiment. The procedures of this study includes
following steps:
 
The first step is the sample collection of errors. This starting point in is deciding what
samples of learner’s language to use for the analysis and how to collect these samples. The
learner’s language should reflect natural and spontaneous language use，or be elicited in
some way.
 
The second step is the identification of errors. Once a corpus of learner’s language has been
collected, the errors in the corpus have to be identified. It is necessary to decide,  therefore,
what constitutes an error and to establish procedure for recognizing one.
 
The third step is the description of errors. There are a number of classification systems that
have been used. In this study，I identify three types of lexical errors. They are spelling
errors，assumed synonymy，mix-up of parts of speech.
 
The fourth step is the explanation of errors. Explanation is concerned with establishing the
source of the errors.  
 
3  Results and Discussion    
3.1 Brief Overview of the Main Results
In this study, 120 compositions are collected  and all of them were regarded valid. The
students are required to write one composition within the limited time. And in those
compositions 652 errors were identified. Among them, 301 errors were in lexis, accounting
for 46.17%. From the statistics, it shows that the number of spelling errors, assumed
synonymy and mix-up of parts of speech are  quite considerable among the 301 lexical
errors. They take the largest proportion of lexical errors, a total of 296, accounting for 98%
in lexical errors. And as in the samples, most of the collocation errors are related to
assumed synonyms. This study will not discuss collocation errors separately, but classify it
into the category of assumed synonyms.
 
3.2 Analysis of Errors in Each Category
3.2.1 Spelling Errors
All the error examples described in this study are taken  from the corpus which the writer
has identified in the students' compositions. There are 69 spelling errors in total. This kind
of errors is very easy to identify. Generally speaking, spelling errors relate to the Chinese
language to some extent, because Chinese is not phonetic writing.  Students may easily
ignore the phonetic forms of English words. The similar feature existing between the target
form and the form used can explain most  spelling errors. However, some performance
errors such as anxiety, bad memory and carelessness are also the causes. For example, I will 非英语专业学生英语写作中的词汇错误分析与对策(4):http://www.youerw.com/yingyu/lunwen_4257.html