Using Computer-Tagged Linguistic Features to Describe L2 Writing Differences

Writing skills and strategies of bilingual immigrant students learning Greek as a second language and English as a foreign language

“…Research on writing (Ascención 2008;Cumming 2001;Grant & Ginther 2000;Stein 2000) indicates that the use of strategies and the knowledge which students bring to the text influence their writing process in a significant way. Furthermore, it has been indicated that problems in the students' approach to writing ultimately reflect a deficiency or lack of awareness of the processes involved in performing a writing task (e.g.…”

Predictive modeling of human placement decisions in an English Writing Placement Test

“…(length of the whole text, third and last paragraph), and some syntactic complexity, although some paragraphs might be under-developed (PC2). The findings of this study partially supported previous research indicating that higher-rated essays showed a variety of syntactic constructions, such as subordination (Grant & Ginther, 2000). In contrast, from Figure 4.15, there was the second group of variables in which the 101B essays seemed to be significantly the highest while the Pass essays were the lowest.…”

“…Several studies (Connor, 1990;Engber, 1995;Ferris, 1994;Frase, Faletti, Ginther & Grant, 1997;Jarvis, 2002;Jarvis, Grant, Bikowski & Ferris, 2003;Reid, 1990;Reppen, 1994) have revealed a range of linguistic features at surface and text-based levels. These features are found to be able to differentiate writing performance levels, as shown in Table 2.2, such as text length (Ferris, 1994;Frase et al, 1997;Grant & Ginther, 2000;Reid, 1990), word types, word length (Ferris, 1994;Frase et al, 1997;Grant & Ginther, 2000;Reid, 1990;Reppen, 1994). The findings suggest that higher-rated essays demonstrate a variety of syntactic constructions and lexical features more frequently, including passives (Connor, 1990;Ferris, 1994;Grant & Ginther, 2000), synonymy/ antonymy (Ferris, 1994), normalizations (Connor, 1990;Grant & Ginther, 2000), subordination, unique word 19 choice (Grant & Ginther, 2000), and personal pronouns (Ferris, 1994).…”

“…However, some methodological limitations that are shared in most accounts of prior studies need to be addressed. The first concern of automatic tagging and analysis, such as Coh-Metrix in NLP-based studies, is raised as a high chance of mistagging or misanalyzing in L2 student essays (Ferris, 1994;Grant & Ginther, 2000). The issue does not necessarily result from reliability of automated machines.…”

“…Therefore, inaccuracy in automated detection possibly distorts analytical results. Therefore, Ferris (1994) and Grant et al (2000) suggest that additional hand-coding is desirable in order to achieve a complete profile of L2 writing proficiency differences. Secondly, unstable results of prior studies in investigating the relationships between text features and essay quality may suffer from methodological classification issues.…”