A Corpus-based Study of Discourse Features in Learners

A Corpus-based Study of Discourse Features in Learners’ Writing Development Yu-Hua Chen Lancaster University, UK

Background Ellis (1994: 44) identified four major approaches in describing learner language: the study of learners’ errors n the study of developmental patterns n the study of variability n the study of pragmatic features n 2

Background Ellis (1994: 44) identified four major approaches in describing learner language: the study of learners’ errors n the study of developmental patterns n the study of variability n the study of pragmatic features n 3

Background n Studies on ESL/EFL writing development 4

Background n Studies on ESL/EFL writing development n Generally focusing on fluency, accuracy and complexity 5

Background Studies on ESL/EFL writing development n Generally focusing on fluency, accuracy and complexity n Usually based on small-scale data 6

Background Studies on ESL/EFL writing development n Generally focusing on fluency, accuracy and complexity n Usually based on small-scale data n Not much attention being paid to the developmental patterns of discourse features 7

My Hypothesis In addition to fluency, accuracy and complexity, there are developmental patterns at the textual level of learners’ written language. 8

Previous Findings For example, Kennedy & Thorp (2007) n IELTS writers at levels 4/6 use lexico-grammatical markers (e. g. however) more frequently. Level 8 writers rely on them less and appear to be nearer to native speaker use in this respect. 9

Previous Findings For example, Kennedy & Thorp (2007) n IELTS writers at levels 4/6 use lexico-grammatical markers (e. g. however) more frequently. Level 8 writers rely on them less and appear to be nearer to native speaker use in this respect. n Enumerative markers (firstly, etc) are used twice as much by level 4/6 writers as level 8 writers. 10

Previous Findings For example, Kennedy & Thorp (2007) n Subordinators are used slightly less by level 8 writers than by levels 4/6. 11

Previous Findings For example, Kennedy & Thorp (2007) n Subordinators are used slightly less by level 8 writers than by levels 4/6. n Lower level scripts (4/6) are more categorical and content is less modified by hedging. Expressions of (dis)agreement and opinion tend to be more directly expressed (I agree/disagree/in my opinion etc) at level 4. 12

Provisional Research Focuses n Bottom-up Approach n Discourse Markers, e. g. however, hence, first, then, etc. 13

Provisional Research Focuses n Bottom-up Approach n n Discourse Markers, e. g. however, hence, first, then, etc. Hedging Devices, e. g. modal auxiliaries, modal lexical verbs (seem, appear, suggest, indicate, assume, etc. ), modal adjectives (possible), modal adverbials (perhaps) 14

Provisional Research Focuses n Bottom-up Approach n n n Discourse Markers, e. g. however, hence, first, then, etc. Hedging Devices, e. g. modal auxiliaries, modal lexical verbs (seem, appear, suggest, indicate, assume, etc. ), modal adjectives (possible), modal adverbials (perhaps)) Top-down Approach n Lexical bundles: recurrent word combination 15

Data n The Longman Learner Corpus A large collection of documents written by learners of English n Comprising mainly essays and exam scripts from students throughout the world between 1990 and 2002 n 16

Data n The Longman Learner Corpus A large collection of documents written by learners of English n Comprising mainly essays and exam scripts from students throughout the world between 1990 and 2002 n n Only written samples from L 1 Chinese learners of L 2 English are included in my project. 17

Data n The Longman Learner Corpus A large collection of documents written by learners of English n Comprising mainly essays and exam scripts from students throughout the world between 1990 and 2002 n Only written samples from L 1 Chinese learners of L 2 English are included in my project. n Only argumentative/expository essays are selected. n 18

A LLC Sample <RF> 00706 <CO> USA <LA> CHC <LE> AD <EN> INT <TT> 1 <PA> Culture shock occurs as a result of when you are suddently immersion in a new different culture. It's due to your own feeling how to deal with the differences in food, clothing, and other things. <PA> When I arrived at the L. A airport on Aug 3. The first thing is to meet the I. N. S. inspector. I found the Americas are very open-minded and kindness to give me more assistance. They judge everyone just built in a Truth. If you are honest, it is easy to pass. Then I woked out the gate of airport I saw it is marvolous that has so many terminals and complex cross-freeway system just a high technology architeture there. 19

The Need to Rate the LLC Scripts It is argued that the judgment on learners’ proficiency should be made based on their linguistic performance, as opposed to ‘extralinguistic factors’ such as years of learning English or programme levels. In addition, most learners are not equally competent in terms of reading, listening, speaking, and writing; therefore, it is necessary to rate the LLC scripts if only the written performance is the target to be analysed.

Rating Scale n the written assessment criteria grid in Manual for relating Language Examinations to the Common European Framework of Reference (p. 82) 21

Rating Scale n the written assessment criteria grid in Manual for relating Language Examinations to the Common European Framework of Reference (p. 82) n Due to the large number of samples needed in this project, a holistic/impressionist rating scale is adopted. 22

Expected Project Scale Level Pilot Study No. of samples Main Investigation No. of words No. of samples No. of words C 2 -- 20, 000 -- 100, 000 C 1 -- 20, 000 -- 100, 000 B 2 -- 20, 000 -- 100, 000 B 1 -- 20, 000 -- 100, 000 Total 300 80, 000 1, 500 400, 000 23

Expected Contributions n Collaboration between SLA and Language Testing 24

Expected Contributions Collaboration between SLA and Language Testing n SLA n Shed light on learners’ language development n Provide insights for ELT writing pedagogy n 25

Expected Contributions Collaboration between SLA and Language Testing n SLA n Shed light on learners’ language development n Provide insights for ELT writing pedagogy n n Language Testing Facilitate the design of a common band scale of writing assessment n Help develop a computer automated marking system n 26