![](bg1.png)
ASSESSING WRITING FLUENCY OF NON-ENGLISH-SPEAKING
STUDENT FOR AUTOMATED ESSAY SCORING
How to Automatically Evaluate the Fluency in English Essay
Min-Chul Yang, Min-Jeong Kim, Hyoung-Gyu Lee and Hae-Chang Rim
Department of Computer and Radio Communications Engineering, Korea University, 136-713, Seoul, South Korea
Keywords:
CBA, Computer-based Assessment, AES, Automated Essay Scoring, Writing, Fluency, Machine Learning.
Abstract:
Automated essay scoring (AES) systems have provided a computer-based writing assessment comparable to
expert raters. However, the existing systems are inadequate to assess the writing fluency of non-English-
speaking students, while they detect grammatical errors relatively well. The writing fluency is one of the
important criteria in essay scoring, because most of non-English-speaking students have much difficulty in
expressing their thoughts in English. In this paper, we propose an automated essay scoring system focusing on
assessing the writing fluency by considering the quantitative factors such as vocabulary, perplexity in a sen-
tence, diversity of sentence structures and grammatical relations. Experimental results show that the proposed
method improves the performance in automated essay scoring.
1 INTRODUCTION
In the era of globalization, the English language has
become the most common language. For this reason,
many people in non-English-speaking countries have
had a great interest in learning English language. En-
glish certification tests such as TOEIC, TOEFL and
GRE are being used to admit the applicants of many
universities or companies. Among the assessment cri-
teria of English, writing has become an important part
due to recent upsurge of interest in practical English,
but assessing the writing ability is the most expensive
and time-consuming activity. Therefore, automated
essay scoring (AES), which is one of the educational
applications of natural language processing, is stand-
ing out as a replacement of expert rater.
Automated essay scoring system is a real-time
system which can automatically evaluate an essay and
give a score and a feedback to the writer of the es-
say without any human efforts. E-rater(Attali and
Burstein, 2006) and IEA(Foltz et al., 1999) have been
used by many students as commercial automated es-
say scoring systems. In particular, e-rater has been ac-
tually utilized in GMAT test. Unfortunately, the core
technology of these commercial systems is not pub-
lished.
The advantages of automated essay scoring sys-
tem are as follows. 1) It can evaluate an essay in real
time. Students can immediately receive ratings for
their essay. 2) It requires low cost. 3) It is easily ac-
cessible over the web, while an expert rater is required
for the manual assessment.
Many studies have found that many automated
essay scoring systems are able to provide an au-
tomatically measured score that is comparable to
human score(Warschauer, 2006; Wang and Brown,
2007). However, they are suitable for assessing essays
of English-speaking students, but not non-English-
speaking students. The main difference between
English-speaking and non-English-speaking students
is the ability to express their thoughts in English. Ac-
tually, English-speaking students may get a high score
of writing if they improve their reasoning skill. How-
ever, non-English-speaking students should learn not
only the reasoning skill but also the ability to express
their thoughts in English. Therefore, for assessing the
writing of non-English-speaking students, the eval-
uation of English expression ability is the essential
criteria. In order to automatically assess it, an auto-
mated essay scoring system should evaluate an essay
by considering the fluency of a sentence or a whole
essay, as well as detecting the grammatical error of a
sentence. Nevertheless, the existing systems have not
fully considered the writing fluency. In other words,
an additional technology for measuring the fluency
is required to precisely evaluate an essay of a non-
English-speaking student.
In this paper, we define the writing fluency as fol-
83
Yang M., Kim M., Lee H. and Rim H..
ASSESSING WRITING FLUENCY OF NON-ENGLISH-SPEAKING STUDENT FOR AUTOMATED ESSAY SCORING - How to Automatically Evaluate
the Fluency in English Essay.
DOI: 10.5220/0003898200830087
In Proceedings of the 4th International Conference on Computer Supported Education (CSEDU-2012), pages 83-87
ISBN: 978-989-8565-07-5
Copyright
c
2012 SCITEPRESS (Science and Technology Publications, Lda.)