O n f e r e n c e

Short paper 
Synthesis writing from multiple sources in a foreign language challenges the writing 
strategies of novice writers. Instead of copy-pasting fragments, students need to 
cognitively engage with the content of the source texts in order to produce an effective 
synthesis (Solé et al., 2013). They also have to broaden the focus of their linguistic 
attention from the word and sentence levels towards the text level in order to make the 
synthesis coherent (Paulus, 1999). Online peer collaboration can help students to master 
these challenges by pooling their cognitive and linguistic resources (Storch, 2005). 
Furthermore, the collaborative task setting fosters content discussion and reader-
orientation (Blin & Appel, 2011). For this collaboration to be effective, students need to 
be scaffolded in the collaborative process (Slavin, 1992).
The presented intervention study sets out to investigate what type of instructional 
support works best for this specific task and a specific target public, namely advanced 
foreign language students in higher education. Two state-of-the-art instructional methods 
to support online collaboration are used: 
scripting
and 
observational learning
(Rummel, 
Spada, & Hauser 2009). Scripting is a well-researched support instrument in Computer 
Supported Collaborative Learning (CSCL) (Fischer, Kollar, Mandl, & Haake 2007). 
Research indicates that the effect of a script on the collaborative process can depend on 
group characteristics (Hämäläinen & Arvaja 2009). Observational learning has been 
shown to foster effective results both for synthesis writing (Raedts, Rijlaarsdam, van 
Waes, & Daems 2007) and for peer revision in a foreign language (Van Steendam, 
Rijlaarsdam, Sercu, & Van den Bergh 2010).
To measure the effect of 
scripting
and 
observational learning
on online synthesis writing 
in higher education, collaborative triads were formed and a three-weeks intervention 
study was set up. In week 1, all students (n=76) received basic training in Google Docs 
communication facilities, but no further instructions on how to proceed with the task. In 
weeks 2 and 3, a cross-sectional design was implemented: in the "script-first" condition 
(n=34), the students followed a collaboration script in week 2 and watched a model 
video of an online collaboration in week 3, and vice-versa in the "model-first" condition 
(n=42). The collaboration process was recorded using the screencast software 
Morae
. All 
recordings were quantitatively analysed for 
efficiency 
(work flow) and interaction 
intensity
(chat and comments). Attitude questionnaires and focus group interviews were 
used to triangulate the observational data. 
To measure the 
efficiency
of the collaborative work flow, the five task steps 
recommended in the model and the script were used as units of analysis. Task steps 
were marked in 
Morae
for each individual recording, and time on task (a built-in function 
of 
Morae
) for all steps was extracted to a database. Comparing the results for each 
session and/or triad, changes in the organisation of the workflow due to instructional 
support and/or due to "learning by doing" can be observed. In order to quantitatively 
measure collaboration 
intensity
, the communication of the groups in two different 
channels was examined: All 
chat turns
were counted and classified according to their 
main focus on (a) work flow, (b) content, or (c) socializing. In addition, all 
comments
on 
the emerging collaborative text were counted and classified according to their focus on 
(a) lower-order concerns (LOCs), that is, discussions about issues concerning word or 
sentence level, such as lexical and grammar problems, or (b) higher-order concerns 
(HOC), that is, discussions on text level. The HOC discussions, in turn, were subdivided 
into content-related issues (HOC1) and issues related to coherence and cohesion (HOC2). 
The results shed new light on the effect of instructional support. (1) Regardless of the 
order of support instruments, scaffolding had a clear bearing on effective work flow 
organization. This was more prominent in the low-proficiency and mixed-proficiency 
triads than in the high-proficiency ones. 

---

-325- 
2014 CALL Conference 
LINGUAPOLIS
www.antwerpcall.be 
(2) Collaboration density increased considerably after watching the model video, 
irrespective of triad composition according to proficiency, and of condition. This might be 
attributed to the fact that the model depicts cues of joint information processing. (3) 
Collaboration density for both LOC and HOC issues increased to a similar extent 
throughout the sessions and in the two conditions. However, HOC1 (content issues) is 
the only focus that still increases considerably in the last sessions discussions, 
independent of the support method used. This indicates a “learning by doing”
-effect as 
there was an enduring increase of attention to content throughout the intervention. 
However, there are some "refractory" triads in the “script
-
first” condition (35% of the