Coherence in L2 Written Discourse

Principal Investigator
Dr Sophia Skoufaki

Investigating coherence in L2 written language

This project investigates coherence in L2 written language. Coherence errors are difficult to identify, so have received little attention from researchers until now.

Dr Skoufaki has explored whether Rhetorical Structure Theory (a theory from computer science) can be usefully applied to locating these errors. Findings will potentially be beneficial to the error-tagging of L2 learner corpora and the design of automated writing evaluation software.

Principal Investigator

The project

This project validates Dr Skoufaki’s approach to coherence error identification and uncovered teachers' definitions of ‘coherence’ and ‘coherence error’ as well as approaches to providing feedback and marking EFL writing for coherence. EFL teachers located coherence errors in the texts that she had already analysed in terms of Rhetorical Structure Theory (RST) and explained what ‘coherence’ meant to them. 

The teachers’ coherence marks and coherence measures calculated based on their annotations were compared with coherence measures calculated based on the output of the independent RST analysis of the same paragraphs. One such coherence measure was coherence break density. Coherence break density of teacher- and RST-identified breaks were correlated significantly and positively for one third of the teachers. Off-topic content and inductive content order were identified as coherence breaks by both the RST analysis and some teachers.

The findings

These findings suggest that language teachers differ in how they define coherence errors and that Rhetorical Structure Theory, an approach which focuses on only one aspect of coherence, cannot predict all the coherence errors identified by EFL teachers.

Key publications

Devising a discourse error tagging system for an English learner corpus

This paper reports on-going research aiming to result in the tagging of discourse errors in the Language Training and Teaching Center English Learner Corpus, a Taiwanese learner corpus of English constructed by the Graduate Institute of Linguistics and the Department of Foreign Languages and Literatures at National Taiwan University, the Department of English at National Chengchi University, and the LTTC, a language testing company.

Read the abstract or download the full text at ResearchGate.

An exploratory application of Rhetorical Structure Theory to detect coherence errors in L2 English writing: Possible implications for automated writing evaluation software

This paper presents an initial attempt to examine whether Rhetorical Structure Theory (RST) (Mann & Thompson, 1988) can be fruitfully applied to the detection of the coherence errors made by Taiwanese low-intermediate learners of English.

View the abstract or download the full text from our research repository.

Get in touch
Centre for Research in Language Development throughout the Lifespan (LaDeLi) University of Essex
Wivenhoe Park, Colchester, CO4 3SQ
Telephone: 01206 872083