Research on Writing Development in the GUGD

Prototypical Performance in Writing by Curricular Level: Stages in L2 Literacy Development

Goal of the project

Analyze development of writing abilities across the curricular levels I – IV. The analysis comprises both a syntactic, more quantitative analysis (using T-units) and a textual (qualitative and quantitative) approach in order to profile developing literacy, particularly at the more advanced levels. Due to the explicit, though general statements that the curriculum makes regarding writing development and due to the explicit genre-based pedagogies in support of such projected writing development, a key interest is to compare these projections with actual performance. The study provides a near unique opportunity to profile language development longitudinally within a curricular context.

The project has received approval from the IRB Board: 03-040C.

The research team

Heidi Byrnes and Hiram Maxim, GUGD, and John Norris, University of Hawai’i
Former and present graduate students: Matthew Adams, Marianna Ryshina, Shana Semler, Ellen Titzkowski
Undergraduate student:Castle Sinicrope (GU undergraduate research assistant)

Note: Due to the rich data that has been collected and will continue to be collected to enable us to examine writing development longitudinally, we are looking forward to future graduate students joining the team as they join the department.


At this point, the project focuses on two sets of data:

  • Prototypical performance writing tasks. These are written by students in all the sequenced courses at the end of each level, typically as the last graded writing assignment. They were created to capture, to the extent that this is possible in one writing task, the abilities that are considered prototypical of learners’ language development in writing at a particular level. They have previously been checked by the practitioners for their ability to elicit the language-oriented features of a curricular level, in terms of discourse level, sentence-level, and lexicogrammatical features.
  • See Developing Writing on this site for further information. Also Heidi Byrnes: “The role of task and task-based assessment in a content-oriented collegiate foreign language curriculum.” Language Testing 19,4:425-443.
  • Base-line writing task. Because of the well-known task effect in L2 performance, the prototypical performance writing tasks are flanked by a base line writing task. In the spring of 2003 and again in spring 2004, students at all curricular levels performed one and the same writing task.This allows analysis of writing development both within task and across the curricular levels.

A third set of data is also being gathered in collaboration with the Free University of Berlin, in cooperation with Professor Norbert Dittma and his assistant, Ralf Heuer-Meuthrath.
Two groups of students will perform the identical task used in the baseline writing task in the GUGD data take: very advanced non-native students studying at the graduate level at the FU, and native speakers.

In their totality the data comprise the full range of abilities, from beginner to native speaker.

Scope of data

At the end of the spring semester 2004, the following files had been collected, coded and analyzed:

1. Prototypical Performance Writing Tasks. A total of 296 writing samples at 4 curricular levels exist. They vary in length from 1-2 double-spaced typed pages in Level I to about 8-10 pages for Text in Context. They thus constitute a particularly rich data set.

2. Base-line writing tasks across the curricular levels: n = 103.

The general approach

Initial syntactic analysis by T-units uses CHILDES as its computer-based coding system. As of October 2003, the first 241 files have been clause coded according to the following codes:
“clause type + C “indicates a coordinated dependent clause where the second subject is or is not elided (e.g., $RELC would be something like “Der Junge, der morgen Fussball spielt $REL und danach lernen muss $RELC”). This contrasts with $IC, which is reserved to indicate coordinated independent clauses.
“clause type + C+” indicates a coordinated dependent clause with a shift in the subject

Data management and coding

Data are stored in a secure environment to which only the researchers have access. Codes were assigned to each student and then to each file for anonymity, parsimoniousness, informational value of the code, and trackability by student.


With data coding and statistical analysis largely completed at the end of the summer 2004, the research team hopes to write up this study for publication during the academic year 2004-2005.


The project is partially funded by a two-year College Curriculum Renewal Grant 2002-2004. Total funding: $ 7,360.– . It is also funded from departmental sources, particularly through graduate summer stipends.

Note: Please follow the “Research Projects, Grants, Collaboration” link on Heidi Byrnes’ web page to see the write-up.

September 3, 2004