You are on page 1of 13

Original article

doi: 10.1111/jcal.12073

bs_bs_banner

A delineation of the cognitive processes


manifested in a social annotation environment
S.C. Li, J.W.C. Pow & W.C. Cheung
Department of Education Studies, Faculty of Social Science, Hong Kong Baptist University, Kowloon Tong, Kowloon, Hong Kong

Abstract This study aims to examine how students’ learning trajectories progress in an online social
annotation environment, and how their cognitive processes and levels of interaction correlate
with their learning outcomes. Three different types of activities (cognitive, metacognitive and
social) were identified in the online environment. The time series of different levels of
cognitive and metacognitive activities correlate with one another, indicating that a kind of
symbiotic or ecological relationship exists among the cognitive and metacognitive processes.
The multiple regression results show that low-cognitive level activities mediate the effect of
high-cognitive level activities on student’s performance in the group inquiry learning project.
The results of multilevel analysis indicate that the average number of highlighted texts posted
per person in a group accounted for over 50% of the variance of the group inquiry learning
project scores. Further, the results also indicate that the level of collaboration within a group
explains over 70% of the variance of the high-level cognitive and metacognitive activities
across different groups. In sum, group collaboration is found to be conducive to fostering
high-level cognitive and metacognitive activities.

Keywords collaborative learning, Diigo, inquiry learning, knowledge building, social annotation, social
bookmarking.

Web technology is drastically changing the ways


Introduction
people acquire information and construct knowledge. It
The advancement in ubiquitous computing has created provides a platform that seamlessly connects private
huge challenges as well as opportunities to education. and public learning spaces across formal and informal
It is believed that the affordance and pervasiveness of contexts (Chan, Roschelle, Hsi, Kinshuk, & Sharples,
technology can transform traditional classroom learn- 2006; Looi et al., 2010; Sharples, Taylor, & Vavoula,
ing to more ubiquitous learning, enabling learners to 2007). As such, there has been a proliferation of studies
acquire right digital content at the right time and at the on the use of Web 2.0 tools to facilitate collaborative
right place (Chen & Li, 2010). Comas-Quinn, learning across contexts, ranging from English vocabu-
Mardomingo, and Valentine (2009); Tan, Lin, Chu, and lary learning (Chen & Li, 2010) to scientific inquiry
Liu (2012); and Traxler (2007) argue that ubiquitous learning (Liu, Peng, Wu, & Lin, 2009a; Liu, Tan, &
Chu, 2009b). Recently, the use of online social anno-
tation to facilitate collaborative learning has received
Accepted: 27 April 2014
considerable attention (Huang, Huang, & Hsieh, 2008;
Correspondence: Sandy C. Li, Department of Education Studies,
Faculty of Social Science, Hong Kong Baptist University, Kowloon Huang, Jeng, & Huang, 2009; Lu & Deng, 2012,
Tong, Kowloon, Hong Kong. Email: sandyli@hkbu.edu.hk 2013). However, most of the earlier studies focus more

© 2014 John Wiley & Sons Ltd Journal of Computer Assisted Learning (2015), 31, 1–13 1
2 S.C. Li et al.

on the architecture of the system (Chao, Chen, & such as social bookmarking and social annotation,
Chang, 2010; Huang et al., 2008; Nokelainen, provide the affordance for promoting high-level cogni-
Miettinen, Kurhila, Floréen, & Tirri, 2005; Wolfe, tive activities. The World Wide Web has become not
2008), with only a few address the issues associated just our largest information repository but an indispen-
with student learning performance (Lin & Lai, 2013; sable platform for exchanging information and
Lu & Deng, 2013; Yang, Yu, & Sun, 2013). Little is working collaboratively, making it the key information
known about the interplay of cognitive, metacognitive source for inquiry and exploratory learning.
as well as other non-cognitive processes manifested in
the systems. In our present study, we aim to examine
Social annotation
how students’ learning trajectories progress in Diigo,
an online social bookmarking and annotation environ- There have been findings showing that annotation can
ment (Nelson, Christopher, & Mims, 2009; Twu, improve students’ performance in reading and writing
2009), and how their cognitive processes and level of (Yang et al., 2013). Marshall (1997) indicates that
interaction correlate with their learning outcomes. Spe- annotation provides a myriad of useful functions in the
cifically, the research questions for this study are three- process of reading. They include procedural signals,
fold: (1) how do students’ cognitive and metacognitive place markings and aids to memory, in situ problem
processes change over time? (2) how do students’ cog- working, records of interpretive activities, traces of
nitive and metacognitive processes manifested in the readers’ attention and incidental reflections. The use of
environment correlate with students’ learning out- annotation can promote deep learning by helping stu-
comes? and (3) how does the level of collaboration dents encode, decode and transform useful input infor-
within a group affect students’ learning outcomes? mation (Nokelainen et al., 2005). However, Chao et al.
Through addressing these three questions, we could (2010) point out that traditional paper-based annotation
deepen our understanding of how the use of an online may pose physical and cognitive barriers to learners
social annotation environment can facilitate collabora- when they want to organize or integrate information
tive inquiry learning. across different printed media. In parallel with the
advancement of Web technology, a lot of work has been
done on the development of online annotation systems
Web 2.0 for learning
(Chao et al., 2010; Glahn, Specht, & Koper, 2009;
In the first Web 2.0 Conference held in 2004, Tim Hoff, Wehling, & Rothkugel, 2009; Hsi, 2007; Huang
O’Reily (2005) outlined some preliminary principles to et al., 2008; Wolfe, 2008). Online annotation systems
help distinguish the first generation from the second can be categorized into two basic classes: the
generation of World Wide Web. As one of the most discussion-oriented system where users can browse
eye-catching principles among others, ‘Harnessing and annotate threaded discussion messages on the
Collective Intelligence’ has become a unique feature of margins; and the document-oriented system where
Web 2.0. With this ‘collective intelligence’ principle users can browse and annotate Web-based documents
(The New Media Consortium and the EDUCAUSE and media (Nokelainen et al., 2005). Most online anno-
Learning Initiative, 2008), Web 2.0 has been developed tation systems provide a central repository where users
in a way to provide more and more collaboration can store, retrieve and collate the annotations they
opportunities in many areas of human development on made in and across different documents. Some of the
an unprecedented scale. This is made possible by the Web-based annotation systems also provide social net-
user-friendly technologies to ‘publish’ people’s knowl- working and threaded discussion facilities that enable
edge, ideas, opinions and comments. As in education, users to share and co-construct annotations with their
we witness the power of Web 2.0’s collective intelli- collaborators (Lu & Deng, 2013; Nelson et al., 2009;
gence with a growing number of opportunities for stu- Twu, 2009; Yang et al., 2013). There exists a body of
dents around the world to learn together through a research studies on online annotation, but most of the
variety of social media and online collaboration plat- studies centre on the system design issues (Chao et al.,
forms. Tu, Blocher, and Roberts (2008) and Collis and 2010; Huang et al., 2008; Nokelainen et al., 2005;
Moonen (2008) also suggest that Web 2.0 technologies, Wolfe, 2008), with a few looking into students’

© 2014 John Wiley & Sons Ltd


Social annotation and learning 3

performance and behaviours in using collaborative orchestrate a multitude of cognitive, metacognitive and
annotations to support their learning (Lin & Lai, 2013; social processes such as employing appropriate strat-
Lu & Deng, 2013; Yang et al., 2013). Most of the egies and modifying one’s behaviours to address the
findings were derived from student perception surveys changing demands of the tasks (Hammouri, 2003;
or quasi-experiments that compared students’ sum- Montague, 1992). While there is an increasing reper-
mative assessment scores between the experimental toire of knowledge about the intricate nature of these
and control groups. Little is known about the cognitive, processes associated with learning (Billing, 2007;
metacognitive and other non-cognitive processes of Greene & Azevedo, 2010; Tsai, 2009; Wolfe & Flewitt,
learning and how these processes are related to the 2010; Yang, 2012), little is known about the relative
activities students are engaged in collaborative annota- contributions of cognitive, metacognitive and social
tion. A more detailed delineation of the students’ learn- processes to learning outcomes when students are
ing outcomes as well as the interplay of these processes engaged in online collaborative inquiry activities. Pre-
would deepen our understanding of the affordance of vious studies indicate that proper use of metacognitive
collaborative annotation. strategies, to a large extent, enhances the likelihood
of success in problem solving (Hammouri, 2003;
Hartman & Everson, 1996). The term ‘metacognition’
Inquiry-based learning
first appears in the early work of Flavell (1976) who
Inquiry-based learning can find its origin in construc- defines it as the self-consciousness of one’s knowledge,
tivist learning theories, such as Dewey (1938), Freire and self-monitoring and self-regulation of one’s own
(1984) and Vygotsky (1978). In his work on Science cognitive processes and outcomes or anything related
education, Schwab (1962) construed inquiry-based to them. Biggs (1985) refers to meta-learners as those
learning as ‘inquiry into inquiry’. Inquiry-based learn- who are aware of their learning goals, task demands
ing falls under the realm of inductive approaches to and personal cognitive resources, and able to deploy
learning which involves a process of exploration appropriate strategies to accomplish the given tasks.
through which learners ask questions, formulate and While cognitive strategies are related to integrating
rigorously test their hypotheses, and make conjectures new knowledge with one’s prior or existing knowledge,
and refutations (Chang & Wang, 2009). Under these metacognitive strategies involve planning and monitor-
premises, learners construct meanings and advance ing reflective and regulatory activities that take place
their understanding through problem solving and before, during and after any cognitive activities (Li,
knowledge synthesis. Wilhelm (2007) identifies a set of Kong, Lee, & Henri, 2006; Yang, 2012). In the context
key elements for a successful inquiry, which include of reading, Yang (2012) expounds that cognitive pro-
(1) defining authentic and ill-structured problem(s); (2) cesses refer to the actions readers take while interacting
working in groups to enable students to develop multi- directly with the text, whereas metacognitive processes
ple perspectives; (3) students’ ownership over the pro- are deliberate, planned strategies by which readers
blems; and (4) setting milestones. Jonassen (1997) regulate and remediate their reading. In his study on
defines well-structured problems as restricted problems how cognitive and metacognitive processes affect
with convergent solutions, and ill-structured problems undergraduate students’ mathematical problem
as those that possess multiple solutions and approaches solving, Hammouri (2003) modified Montague’s
to solutions. Ill-structured problems usually contain (1992) coding scheme and defined cognitive processes
ambiguity about which principles, rules or theories are as (1) read to comprehend the problem situation; (2)
essential for the solution. Thus, well-crafted inquiry- paraphrase the problem in one’s own words; (3) visu-
based learning is instrumental in fostering students’ alize the problem with mental images; (4) make
high-level cognitive and metacognitive abilities. hypotheses; (5) make conjectures and predictions; and
(6) evaluate the validity of solutions. To understand
students’ online interaction, Song and McNary (2011)
Cognitive, metacognitive and social processes
encode different levels of cognitive processes with the
Meaningful inquiry learning requires less structured keywords such as agree/disagree, inform, elaborate,
problems through which learners are expected to argue, explain, summarize and infer. Apart from

© 2014 John Wiley & Sons Ltd


4 S.C. Li et al.

cognitive and metacognitive processes, the literature on five groups, with each group comprising four to six
online collaborative learning indicates that successful members.
online problem solving is also linked to a wide range of
non-cognitive factors such as mutual trust, accountabil-
Pedagogical design
ity and social support (Hung, Lee, & Lim, 2012; Soller,
2001; Yang, 2012). To address the social dimension To enhance students’ ownership over the inquiry, each
of online learning, Song and McNary (2011) use ‘moti- group was required to research into a chosen social
vate’, ‘appreciate’, ‘encourage’ and ‘reinforce’ to issue. To facilitate meaningful learning, feedback from
encode behaviours that acknowledge peer actions the teacher was provided to each group to ensure that
within a group. Hammouri (2003) asserts that the three their chosen issue was authentic, complex and ill struc-
types of processes interlink with one another and that tured. The chosen topics range from social and eco-
metacognitive processes provide the impetus forces nomic issues to environmental issues, for instance: (1)
which influence both cognitive and social behaviours. Does globalization bring about prosperity or poverty?
Interestingly, it was found in some studies that (2) Should the right to public knowledge override the
metacognitive processes showed no ‘direct’ effect on right to privacy or vice versa? (3) Is the currently pro-
student learning outcomes, yet they produced signifi- posed moral and national education an angel or a devil
cant ‘indirect’ effects on learning by means of cog- to the Hong Kong society? (4) Should urban develop-
nitive processes (Purpura, 1999; Yang, 2012). This sug- ment override environmental conservation or vice
gests that further work to unfold the interaction among versa? Each group was required to co-author a report
these processes is deemed necessary. on the chosen issue to explicate the perspectives of
different stakeholders as well as the collective views of
the group with arguments and supporting evidence. To
Methods facilitate collaborative inquiry, each group was
required to use Diigo to share their bookmarks, high-
Context and participants
light and tag relevant texts, make annotations with
The present study scrutinizes how online social anno- sticky notes, and co-construct argumentation wherever
tation facilitates group-based inquiry learning. The appropriate. Diigo provides students with a toolbar to
context of the study was a university-level course on plug into their Web browsers across different platforms
technology in education delivered with the support of such as desktop computers, tablets and smartphones.
Moodle, an open-source course management system Students can bookmark Web pages, highlight texts or
adopted by the university concerned. This is one of the images, and make annotation with sticky notes while
core foundation courses of a teacher education pro- surfing the Web with their mobile devices. As shown in
gramme with an aim to develop in students the essen- Figure 1, each sticky note is associated with a discus-
tial theories and pedagogical skills for integrating sion forum where they can build on the artefacts
technology in teaching and learning. The course cur- created by their collaborators. For better organization,
riculum consists of an assessment component which retrieval and sorting of information, the system allows
required students to engage themselves in a collabora- students to assign a set of tags or keywords to each
tive inquiry-based learning project. The objective of bookmark. Students can also subscribe to Diigo’s Rich
this group project is to enable students to develop first- Site Summary service to keep themselves informed of
hand experience in using technology to support col- other group members’ postings on the system.
laborative learning, and to help them identify and The entire project was divided into four different
reflect on the key issues arising from the process of stages: Stage 1 – initialization, Stage 2 – exploration,
learning. Two classes on this foundation course were Stage 3 – consolidation, Stage 4 – celebration. At Stage
offered respectively in the first and second semesters 1, the teacher helped students identify a social issue of
during the academic year 2012–2013, with each class interest, define each member’s role, set goals and mile-
comprising 24 students. Thus, a convenient sample of a stones for their learning, familiarize themselves with
total of 48 students (14 male and 34 female) was the social annotation system, and set guidelines for
involved in this study. Each class was divided into group collaboration. At Stage 2, each member was

© 2014 John Wiley & Sons Ltd


Social annotation and learning 5

Figure 1 An Example of an Annotated Website with Highlight Texts and Sticky Notes

expected to contribute at least a minimum number of against the coding scheme given in Table 1. The unit of
relevant websites with annotations, and to respond to or analysis was a student’s posting on Diigo, which could
cross-annotate at least one member’s postings each be a bookmark, a highlighted text or a sticky note.
week. At Stage 3, students were required to collate the To evaluate the quality of each posting, we adopted
information they solicited as well as the annotations Kim and Lee’s (2012) approach by attaching a three-
and the content of their discussions. Based on the dis- level Quality Score (QS) labelled zero, one and two
tilled information, they were expected to construct their according to its levels of relevance and accuracy. The
own mind maps and develop their arguments. At the posting was then further categorized and assigned
final stage, each group was invited to present their a Weighted Score according to its cognitive or
findings and collective views and arguments over the metacognitive levels given in Table 1. The data were
issue with supporting facts and evidence.
Table 1. Coding Scheme for Analysing Different Levels of
Activities Students were Engaged in Diigo

Data and coding Weighted


Dimension Label Code Score
Students’ bookmarks, highlighted texts, sticky notes,
discussion postings and their project scores were the Cognitive1 c01 Agree 1
data sources for the study. As suggested by Hammouri c02 Inform 1
c03 Elaborate 1
(2003), Purpura (1999) and Yang (2012), cognitive and c04 Classify 1
metacognitive processes have respectively direct and c05 Illustrate 1
indirect effects on student learning outcomes. To Cognitive2 c06 Question 3
c07 Criticize 3
unfold the interaction among these processes, we
c08 Summarize 5
modified Song and McNary’s (2011) coding scheme to c09 Synthesize 5
incorporate a metacognitive dimension into the frame- c10 Evaluate 5
work as shown in Table 1. As proposed by Li et al. Metacognitive m01 Reflect 7
m02 Manage 7
(2006), the metacognitive dimension encompasses m03 Plan 7
reflection and self-regulation, planning, and manage- Social s01 Appreciate –
ment of students’ own learning. All the highlighted s02 Request –
s03 Encourage –
texts, sticky notes and discussion postings were coded

© 2014 John Wiley & Sons Ltd


6 S.C. Li et al.

input into a spreadsheet manually and coded by one postings contributed by members within a group. In
researcher and one research assistant with an inter-rater this case, a smaller sd’ value indicates that the number
reliability of above 0.7. The Aggregate Score of each of posted items is more evenly distributed among
posted item was defined as the product of its QS and members of a group. As given in Table 2, Group A had
Weighted Score. the largest sd’ value, signifying that the posted items
To delineate the progression of students’ cognitive were contributed mostly by a few group members. To
processes, we computed three scores (Cognitive1 gauge the level of collaboration within a group, we
Score, Cognitive2 Score, Metacognitive Score) to char- computed a new quantity, Collab, by dividing M by
acterize different levels and types of activities identi- sd’. A high score of Collab implies that the group
fied in their inquiry learning. For each student, we then activities in Diigo are intensive (with a large M value),
computed a Cognitive1 Score by adding the Aggregate and/or each group member has a fair contribution to the
Scores of all the items posted by the student under the activities (a low sd’ value). As such, Group A and
Cognitive1 dimension. Likewise, Cognitive2 Score and Group F exhibited a high level of collaboration as com-
Metacognitive Score were computed in the same pared to the other groups (see Table 2). ANOVA was
fashion to reflect the levels of students’ engagement in also conducted on the total number of posted items
the collaborative inquiry learning. against gender, and no statistically significant differ-
ence was found between the male and female groups.
Results To delineate the levels of activities students were
engaged in, all posted items including bookmarks,
Students’ activity profiles in Diigo environment
highlighted texts and sticky notes were coded with
The numbers of bookmarks, highlighted texts, sticky respect to the scheme given in Table 1. The cognitive
notes and discussion threads posted by each group dimension is divided into ‘Cognitive1’ referring to
were depicted in Table 2. The average number of Level-1 cognitive processes such as ‘agree’, ‘disagree’,
posted items per person in each group (M) ranged from ‘inform’, ‘classify’ and ‘elaborate’, and ‘Cognitive2’
7.8 to 72, and the standard deviation (sd) varied from referring to processes of a higher cognitive level such
5.50 to 46.1. The average number of posted items, M, as ‘question’, ‘criticize’, ‘summarize’, ‘synthesize’
can be used as a measure to gauge the intensity of each and ‘evaluate’. The processes associated with the
group’s activities in Diigo. Based on the M values metacognitive dimension were categorized into three
given in Table 2, Group F and Group G were respec- groups, namely ‘reflect’, ‘manage’ and ‘plan’, while
tively the most active and inert groups. To better illus- the social processes were coded with the keywords:
trate the variation in members’ contribution in each ‘appreciate’, ‘request’ and ‘encourage’. With the
group, we defined the normalized standard deviation, scheme given in Table 1, most of the text-highlighting
sd’, as the sd divided by the total number of posted activities were coded under ‘Cognitive1’. A sample of
items. sd’ thus represents the sd of the fraction of sticky notes posted by students is given in Table 3. The

Table 2. The Mean and Standard Deviation of Each Group’s Postings in Diigo

Highlighted Sticky Discussion Collab


Group Bookmark text note thread Total M SD SD’ (M/SD’)

A 41 91 96 0 228 45.6 46.1 0.202 225.4


B 9 13 20 1 43 8.6 4.31 0.102 84.12
C 28 41 76 1 146 29.2 17.1 0.117 248.7
D 20 65 52 0 137 34.3 10.5 0.077 446.8
E 31 64 65 0 160 32.0 13.8 0.087 368.5
F 29 100 159 0 288 72.0 24.9 0.087 829.6
G 11 10 10 0 31 7.8 5.50 0.179 43.25
H 16 140 36 0 192 38.4 38.7 0.202 190.2
I 24 104 27 0 155 31.0 14.5 0.094 329.2
J 10 66 33 0 109 18.2 8.43 0.078 233.7

SD’ = normalized standard deviation.

© 2014 John Wiley & Sons Ltd


Social annotation and learning 7

Table 3. A Sample of Sticky Notes Posted by Students

Dimension Code Sticky note

Cognitive1 Agree Agreed! The radioactive waste disposal is a big problem for our next generation!
Disagree I don’t agree with Kelvin! Wiki only provides information, all you need is critical
thinking to identify the information is fake or not.
Inform For more information about nuclear disaster, please refer to the website’s pdf file.
Elaborate Wow! I see, nuclear power is not clean and 100% environmental, the waste cannot be
destroyed and, it may leak out and cause troubles.
Classify . . . these refer to problems arising from GM food . . . .
Illustrate WHO claimed that ‘GM foods currently available on the international market have
passed risk assessments and are not likely to present risks to human health.’
Cognitive2 Question Can students develop their critical thinking if the national education curriculum is
implemented? It really depends on the teaching materials and teachers’ standpoints.
Criticize The views presented in this website cannot represent the opinions of the majority of the
residents living at Tai Po. We even don’t know the background of this society and its
composition. The author doesn’t indicate the sample size for their opinion poll.
Summarize To sum up, the development of north east areas in Hong Kong would probably bring
lots of benefits according to writer’s opinions: (1) increase the residential uses to
satisfy the increasing demand of population; (2) promote the development of
high-tech industry . . . and (3) promote the long-term economic development . . . .
Synthesize According to the statistics provided, we can see that the development of north east area
of Hong Kong can largely increase the housing supply, both in public and private
housing, which can greatly reduce the existing pressure on accommodation.
Evaluate Excellent article from the Conservation Group of Northeast New Territories. It provided
their detailed reasons to oppose the development plan. More importantly, suggestions
were given for further improvement of the plan. Please read it all of you!
Metacognitive Reflect It’s a kind of an irony that it is more likely to let student become less capable to
distinguish right from wrong as some wrong information or information that had not
been proven right is included in the curriculum.
Manage We need to have a good division of labour. Apart from working on your assigned area,
please also forward the information to us whenever you come across useful websites.
Plan The words no. should be about 2000 words, so I think the background of the issue
should be 400 words (250 words include Hong Kong context), . . . I will first make a
draft about the background of the issue, than you all make some comment or edit,
ok?
Social Appreciate Thanks to John. It is the official document that I am looking for! It is of great help to
me . . . . . . .
Request Sorry that I cannot open the link of the website! Can you check this link for me?
Encourage It is great! . . . . The pictures are well presented and I do like that! Well done!

figures given in Table 4 indicate that most of the activ- with each other. The more Cognitive1 activities stu-
ities in Diigo were of ‘Cognitive1’ type. As the level dents were engaged in, the more Cognitive2 and
of cognitive activities escalated, the number of posted Metacognitive activities were recorded in Diigo. As
items decreased. The numbers of posted items relating mentioned above, Cognitive1 activities comprised
to metacognitive and social activities were small as mainly of text-highlighting activities. Thus, the
compared to other types of activities. As 5 out of the 10 results depicted in Figure 2 seem to indicate that text-
groups showed little evidence of social activities, our highlighting activities help provide a context
discussion will mainly focus on activities associated for students to develop high-level cognitive and
with Cognitive1, Cognitive2 and Metacognitive metacognitive activities. To further elaborate the rela-
dimensions. tionship among the three types of activities over time,
Figure 2 shows the variation of different types of we used IBM SPSS 21 package to examine the cross-
activities over time. It was evident that the number of correlation of the three time series, Cognitive1 Activ-
Cognitive1, Cognitive2 and Metacognitive activities ities (T_C1), Cognitive2 Activities (T_C2) and
recorded during a day had a strong association Metacognitive Activities (T_MC), across different

© 2014 John Wiley & Sons Ltd


8 S.C. Li et al.

Table 4. Numbers of Cognitive1,


Dimension Cognitive2, Metacognitive and Social
Activities Identified in Each Group
Group Cognitive1 Cognitive2 Metacognitive Social Total

A 172 28 5 23 228
B 27 9 1 6 43
C 102 33 11 0 146
D 108 25 3 1 137
E 115 33 1 11 160
F 161 102 11 14 288
G 24 7 0 0 31
H 165 26 1 0 192
I 143 8 4 0 155
J 86 22 1 0 109

groups. The results presented in Table 5 indicate that


Multiple regression
the three types of activities interweaved with one
another: T_C1 correlated strongly with T_C2 and To quantify and differentiate different levels of activ-
T_MC across different groups, except for Group G, as ities, we computed for each student three scores, Cog-
no metacognitive activity was identified in the group. nitive1 Score (C1), Cognitive2 Score (C2) and
In other words, the more Cognitive1 activities a group Metacognitive Score (MC), by multiplying each posted
was engaged in during a day, the more Cognitive2 and item’s cognitive level score with its QS. The Total Score
Metacognitive activities were promoted. (TS) was calculated by adding Cognitive1, Cognitive2

Figure 2 Variation of the Numbers of Cognitive1, Cognitive2 and Metacognitive Activities Over Time

© 2014 John Wiley & Sons Ltd


Social annotation and learning 9

Table 5. Cross-Correlation of the Time Series for Cognitive1, Cognitive2 and Metacognitive Activities Across Different Groups

Group

A B C D E F G H I J

r1 .816** 1.0** .70** .97** .75** .88** .66* .93** .93** .99**
r2 .300** 1.0** .36** .78** .51** .61** – .27* .54** .95**

r1 = cross-correlation coefficients between T_C1 and T_C2. r2 = cross-correlation coefficients between T_C1 and T_MC.
*p < 0.05; **p < 0.01.

and Metacognitive Score together. The results given in with IBM SPSS 21 package. Four regression models
Table 6 indicate that the numbers of bookmarks (BM), were depicted in Table 7. The R2 values given in Model
highlighted texts (HL) and sticky notes (SN) posted by 1, Model 2 and Model 3 indicate that TS, Weighted
each student strongly correlated with each other. Book- Cognitive1 Score and Cognitive2 Score are strongly
marks, highlighted texts and sticky notes also have associated with PS, explaining respectively 14%, 15%
strong correlation with TS, Cognitive1, Cognitive2 and and 8% of the variance in PS. In Model 4, Cognitive1
Metacognitive Score. In addition, highlighted texts, and Cognitive2 were put together as the two independ-
sticky notes, Cognitive1, Cognitive2 and TS correlated ent variables for the analysis. The results indicate that
strongly with Project Score (PS), while the correlation the regression coefficient for Cognitive2 is statistically
between bookmarks and PS, and between Metaco- insignificant. After controlling for the effect of Cogni-
gnitive Score and PS, was not statistically significant. tive1 on PS, the % variance of PS explained by Cog-
To further examine the interplay among students’ nitive2 drops from 8% to 3.7% (16.8%–15.1%),
different levels of cognitive activities and their project indicating that Cognitive1 mediates the association
scores, multiple regression analysis was conducted between Cognitive2 and PS.

Table 6. Correlation Among Different Types of Activities Students Engaged in and Their Project Scores

Correlation coefficient

BM HL SN TS C1 C2 MC PS

Bookmark (BM) 1 .456** .620** .603** .728** .351* .508** .254


Highlight (HL) .456** 1 .462** .597** .898** .307* .305* .348*
Sticky note (SN) .620** .462** 1 .934** .701** .843** .594** .316*
Total Score (TS) .603** .597** .934** 1 .748** .898** .654** .370**
Cognitive1 Score (C1) .728** .898** .701** .748** 1 .424** .486** .389**
Cognitive2 Score (C2) .351* .307* .843** .898** .424** 1 .415** .283*
Metacognitive Score (MC) .508** .305* .594** .654** .486** .415** 1 .196
Project Score (PS) .254 .348* .316* .370** .389** .283* .196 1

*p < 0.05 level (two tailed); **p < 0.01 level (two tailed).

Table 7. Multiple Regression Models with


Project Score as the Dependent Variable Regression (Project Score as dependent variable)

Model Independent variable β t R2 F

1 Total Score (TS) .370* 2.70** .137** 7.30**


2 Cognitive1 Score (C1) .389** 2.86** .151** 8.20**
3 Cognitive2 Score (C2) .283* 2.00* .080* 4.01*
4 Cognitive1 Score (C1) .328* 2.19* .168* 4.55*
Cognitive2 Score (C2) .144 0.96

*p < 0.05; **p < 0.01.

© 2014 John Wiley & Sons Ltd


10 S.C. Li et al.

Table 8. Two-Level Models for Examining


Variance component the Group Effects on Project Score,
Model Description γ01 of u0j P-value Cognitive2 Score and Metacognitive Score

A0 (PS)ij = β0j + rij – 7.62 <0.001


β0j = γ00 + u0j
A1 (PS)ij = β0j + rij 0.05 3.75 0.003
β0j = γ00 + γ01 (G_HL)j + u0j
B0 (WC2)ij = β0j + rij – 4160 0.020
β0j = γ00 + u0j
B1 (WC2)ij = β0j + rij 0.26 972 0.002
β0j = γ00 + γ01 (Collab)j + u0j
C0 (WMC)ij = β0j + rij – 95.2 0.019
β0j = γ00 + u0j
C1 (WMC)ij = β0j + rij 0.04 28.0 <0.001
β0j = γ00 + γ01(Collab)j + u0j

results were obtained, and the % variance of (PS)ij


Multilevel analysis
explained by the predictors is negligible. In Model B1,
Three sets of two-level models (A0, A1, B0, B1, C0, C1) each group’s level of collaboration score, (Collab)j, is
were constructed to examine the group effect of the chosen as the Level-2 predictor. The % difference
Cognitive1 Score, Cognitive2 Score, Metacognitive between the unconditional variance in Cognitive2
Score and PS using HLM 7 (Scientific Software Inter- Score, (C2)ij, over groups and variance in (C2)ij over
national) package. As shown in Table 8, Model A0, groups after each group’s (Collab)j has been taken into
Model B0 and Model C0 are the fully unconditional account is equal to 76.6% ([4160 – 972]/4160*100%).
models in which PS, Cognitive2 Score and Likewise, each group’s level of collaboration score,
Metacognitive Score are the respective outcome vari- (Collab)j, is also chosen as the Level-2 predictor for
ables, and no Level-1 and Level-2 predictors are Model C1. The % difference between the unconditional
chosen. In these fully unconditional models, the vari- variance in Metacognitive Score, (MC)ij, over groups
ance of the outcome variable in each model is divided and variance in (MC)ij over groups after each group’s
into two components: the within-group variance (rij) (Collab)j has been taken into account is equal to 70.6%
and the between-group variance (u0j). Models A1, B1 ([95.2 – 28]/95.2*100%). Thus, the level of collabora-
and C1 are models with no Level-1 predictors but a tion of each group accounted for the variation in
single Level-2 predictor. The Level-2 predictor is used each group’s high-level cognitive and metacognitive
to examine how much the unconditional between- activities. A similar two-level model was constructed
group variance can be explained by the model. In to examine the group-level effect on Cognitive1 Score.
Model A1, the total number of highlighted texts posted However, no statistically significant result was derived.
by each group (G_HL)j is chosen as the Level-2 pre-
dictor. The % difference between the unconditional
Discussion and conclusion
variance in Project Score, (PS)ij, over groups and vari-
ance in (PS)ij over groups after each group’s total The results presented above deepen our understanding
number of highlighted texts posted has been taken into of how the use of online social annotation facilitates
account is equal to 50.7% ([7.62 – 3.75]/7.62*100%). collaborative inquiry learning. Three different types of
Thus, the average number of highlighted texts posted activities, namely cognitive, metacognitive and social,
per person in a group, (G_HL)j, has a high predictive were identified in the online environment. The time
power over the variation in group-project performance series of the Cognitive1 activities correlates strongly
among the groups. Similar models were constructed with those of the Cognitive2 and Metacognitive activ-
using the total number of bookmarks and the total ities, indicating that a kind of symbiotic or ecological
number of sticky notes posted by each group as the relationship exists among the three types of activities.
Level-2 predictors, but no statistically significant The more Cognitive1 activities students were engaged

© 2014 John Wiley & Sons Ltd


Social annotation and learning 11

in during a day, the more Cognitive2 and metacognitive accounts for most of the variation of Cognitive2 Score
activities were fostered in a group. The numbers of the and Metacognitive Score across the groups, suggesting
three types of activities formed a pyramid similar to that active participation and a distributive share of con-
the population pyramid found in an ecosystem, with the tribution among group members help promote high
number of Cognitive1 activities at the bottom and the levels of cognitive and metacognitive activities. This
number of metacognitive activities at the top. The cor- finding echoes with the notion of social constructivism
relation analysis indicates that the Cognitive1 activ- that culture and the social context is of critical impor-
ities, such as highlighting text, correlate with higher tance for cognitive development. It is evident that the
level of activities and also student’s project score. That online social annotation environment provides the
is, the more highlighted texts a student posted, the more affordance for promoting high-level cognitive and
Cognitive2 and Metacognitive activities he/she metacognitive activities. In particular, the text-
engaged in, and the higher project score he/she highlighting facility not only levers learners’ compre-
attained. The multiple regression analysis also shows hension ability but also provides the necessary anchors
that Cognitive1 activities mediate the effect of Cogni- and scaffolds for learners to engage in meaning nego-
tive2 activities on student’s performance in the group tiations and knowledge building.
project. The effect of Metacognitive activities was While discussing the affordance of the social anno-
found to be statistically not significant. The results tation system, we should not play down the role of
echo the findings derived from the studies conducted pedagogical and social factors in engendering mean-
by Hammouri (2003), Purpura (1999) and Yang ingful inquiry learning experiences. Obviously, tech-
(2012), that metacognitive processes have only indirect nology per se may not suffice to boost learning
effect on student learning outcomes. It could be attrib- motivation nor foster high-level cognitive and
uted to the fact that the aggregated number of metacognitive processes. The pedagogical factors such
metacognitive activities was relatively small as com- as the authenticity and the ill-structured nature of prob-
pared to the number of the cognitive activities, so it was lems, students’ ownership over the problems, and teac-
not sufficient to account for the variation in students’ her’s facilitation remain crucial for designing inquiry
project scores. To look further into the effect of learning and engaging learners.
metacognitive activities on students’ learning out-
comes, a longitudinal study with a larger sample size
Limitations and suggestions
and a longer project span is required.
The effect of text-highlighting activities on student The purpose of this research is not to make any gener-
learning was further elaborated with the results of the alized claims on the use of online social annotation
multilevel analysis. It was shown that the average environment to support learning. Thus, the results of
number of highlighted texts posted per person in a group our study should be interpreted as illustrative rather
accounted for most of the variation of project scores than definitive. Although the use of online social anno-
across the groups, indicating that text-highlighting tation environment seems to have a positive effect on
helped enhance group inquiry and student learning out- student learning outcomes, yet our knowledge of what
comes. Our interpretation is that the highlight function factors are conducive to fostering collaboration and
provided by the online annotation environment helps different levels of cognitive and metacognitive activ-
students to have a more in-depth comprehension of ities is still limited. Furthermore, a more detailed
information and the texts highlighted provide a rich analysis of the time series by shortening the time inter-
context for students to engage themselves in further val for event recording from 1 day to an hour or even a
group deliberations and high levels of cognitive and minute will deepen our understanding of the interplay
metacognitive activities. Our findings contrast with the of different levels of cognitive and metacognitive activ-
results of some studies on traditional discussion boards ities. Last but not least, the pedagogical factors such as
(Nisbet, 2004; Song & McNary, 2011) that students’ teacher’s facilitation and the pedagogical design
postings are of little value and did not contribute to warrant our attention. To shed light on the above issues,
learning. The results of the multilevel analysis further a longitudinal study with in-depth interviews for both
signify that the level of collaboration within a group teacher and students is deemed necessary.

© 2014 John Wiley & Sons Ltd


12 S.C. Li et al.

Hartman, H., & Everson, H. (1996). Self-concept and


References metacognition in ethnic minorities. Urban Education, 31,
Biggs, J. B. (1985). The role of metalearning in study pro- 222–237.
cesses. The British Journal of Educational Psychology, 55, Hoff, C., Wehling, U., & Rothkugel, S. (2009). From paper-
185–212. and-pen annotations to artefact-based mobile learning.
Billing, D. (2007). Teaching for transfer of core/key skills in Journal of Computer Assisted Learning, 25, 219–237.
higher education: Cognitive skills. Higher Education, 53, Hsi, S. (2007). Conceptualizing learning from the everyday
483–516. activities of digital kids. International Journal of Science
Chan, T.-W., Roschelle, J., Hsi, S., Kinshuk, K., & Sharples, Education, 29, 1509–1529.
M. (2006). One-to-one technology enhanced learning: An Huang, Y.-M., Huang, T.-C., & Hsieh, M.-Y. (2008). Using
opportunity for global research collaboration. Research annotation services in a ubiquitous Jigsaw cooperative
and Practice of Technology Enhanced Learning, 1, learning environment. Educational Technology & Society,
3–29. 11, 3–15.
Chang, C.-Y., & Wang, H.-C. (2009). Issues of inquiry learn- Huang, Y.-M., Jeng, Y.-L., & Huang, T.-C. (2009). An edu-
ing in digital learning environments. British Journal of cational mobile blogging system for supporting collabora-
Educational Technology, 40, 169–173. tive learning. Journal of Educational Technology &
Chao, P.-Y., Chen, G.-D., & Chang, C.-W. (2010). Develop- Society, 12, 163–175.
ing a cross-media system to facilitate question-driven Hung, D., Lee, S.-S., & Lim, K. Y. T. (2012). Authenticity in
digital annotations on paper textbooks. Educational Tech- learning for the twenty-first century: Bridging the formal
nology & Society, 13, 38–49. and the informal. Educational Technology Research and
Chen, C.-M., & Li, Y.-L. (2010). Personalised context-aware Development, 60, 1071–1091.
ubiquitous learning system for supporting effective Jonassen, D. H. (1997). Instructional design models for
English vocabulary learning. Interactive Learning Envi- well-structured and ill-structured problem-solving learn-
ronments, 18, 341–364. ing outcomes. Educational Technology: Research and
Collis, B., & Moonen, J. (2008). Web 2.0 tools and processes Development, 45, 65–95.
in higher education: Quality perspectives. Educational Kim, M., & Lee, E. (2012). A multidimensional analysis tool
Media International, 45, 93–106. for visualizing online interactions. Educational Technol-
Comas-Quinn, A., Mardomingo, R., & Valentine, C. (2009). ogy & Society, 15, 89–102.
Mobile blogs in language learning: Making the most of Li, S. C., Kong, S. C., Lee, F. L., & Henri, J. (2006). Capacity
informal and situated learning opportunities. ReCALL, 21, building for lifelong learning: A study of practitioners’
96–112. perceptions on information literacy. Informatics in Educa-
Dewey, J. (1938). Logic: the theory of inquiry. New York, tion, 5, 231–244.
NY: Rinehart and Winston. Lin, J.-W., & Lai, Y.-C. (2013). Harnessing collaborative
Flavell, J. H. (1976). Metacognitive aspects of problem annotations on online formative assessments. Educational
solving. In L. B. Resnick (Ed.), The nature of intelligence Technology & Society, 16, 263–274.
(pp. 231–235). Hillsdale, NJ: Erlbaum. Liu, T.-C., Peng, H., Wu, W.-H., & Lin, M.-S. (2009a). The
Freire, P. (1984). Pedagogy of the oppressed. New York, NY: effects of mobile natural-science learning based on the 5E
Continuum Publishing Company. learning cycle: A case study. Journal of Educational Tech-
Glahn, C., Specht, M., & Koper, R. (2009). Visualisation nology & Society, 12, 344–358.
of interaction footprints for engagement in online Liu, T.-Y., Tan, T.-H., & Chu, Y.-L. (2009b). Outdoor natural
communities. Educational Technology & Society, 12, science learning with an RFID-supported immersive
44–57. ubiquitous learning environment. Journal of Educational
Greene, J. A., & Azevedo, R. (2010). The measurement of Technology & Society, 12, 161–175.
learners’ self-regulated cognitive and metacognitive pro- Looi, C.-K., Seow, P., Zhang, B., So, H.-J., Chen, W., &
cesses while using computer-based learning environments. Wong, L.-H. (2010). Leveraging mobile technology for
Educational Psychologist, 45, 203–209. sustainable seamless learning: A research agenda. British
Hammouri, H. A. M. (2003). An investigation of undergradu- Journal of Educational Technology, 41, 154–169.
ates’ transformational problem solving strategies: Lu, J., & Deng, L. (2012). Reading actively online: An
Cognitive/metacognitive processes as predictors of exploratory investigation of online annotation tools for
holistic/analytic strategies. Assessment & Evaluation in inquiry learning. Canadian Journal of Learning and Tech-
Higher Education, 28, 571–586. nology, 38, 1–16.

© 2014 John Wiley & Sons Ltd


Social annotation and learning 13

Lu, J., & Deng, L. (2013). Examining students’ use of online Tan, T.-H., Lin, M.-S., Chu, Y.-L., & Liu, T.-Y. (2012). Edu-
annotation tools in support of argumentative reading. cational affordances of a ubiquitous learning environment
Australasian Journal of Educational Technology, 29, 161– in a natural science course. Educational Technology &
171. Society, 15, 206–219.
Marshall, C. C. (1997) Annotation: From paper books to the The New Media Consortium and the EDUCAUSE Learning
digital library. Retrieved from http://www.csdl.tamu.edu/ Initiative. (2008) The Horizon Report 2008. 2008 ed. CA.
∼marshall/dl97.pdf (last accessed 26 February 2014). Traxler, J. (2007). Defining, discussing and evaluating
Montague, M. (1992). The effect of cognitive and mobile learning: The moving finger writes and having writ.
metacognitive strategy instruction on the mathematical International Review of Research in Open and Distance
problem solving of middle school students with learning Learning, 8, 1–12.
disabilities. Journal of Learning Disabilities, 25, 230– Tsai, M.-J. (2009). The model of strategic e-Learning:
248. Understanding and evaluating student e-Learning from
Nelson, J., Christopher, A., & Mims, C. (2009). TPACK and metacognitive perspectives. Educational Technology &
Web 2.0: Transformation of Teaching and Learning. Society, 12, 34–48.
TechTrends: Linking Research & Practice to Improve Tu, C.-H., Blocher, M., & Roberts, G. (2008). Constructs for
Learning, 53, 80–85. Web 2.0 learning environments: A theatrical metaphor.
Nisbet, D. (2004). Measuring the quantity and quality of Educational Media International, 45, 253–269.
online discussion group interaction. Journal of eLiteracy, Twu, H.-L. (2009). Effective wiki strategies to support high-
1, 122–139. context culture learners. TechTrends: Linking Research &
Nokelainen, P., Miettinen, M., Kurhila, J., Floréen, P., & Practice to Improve Learning, 53, 16–22.
Tirri, H. (2005). A shared document-based annotation tool Vygotsky, L. S. (1978). Mind in society: The development of
to support learner-centred collaborative learning. British higher psychological processes. Cambridge, MA: Harvard
Journal of Educational Technology, 36, 757–770. University Press.
O’Reily, T. (2005) What is Web 2.0? O’Reilly. Retrieved Wilhelm, J. D. (2007). Engaging readers & writers with
from http://oreilly.com/web2/archive/what-is-web-20.html inquiry (theory and practice). New York, NY: Scholastic
(last accessed 4 September). Inc.
Purpura, J. E. (1999). Learner strategy use and performance Wolfe, J. (2008). Annotations and the collaborative digital
on language tests: A structural equation modeling library: Effects of an aligned annotation interface on
approach. Cambridge: University of Cambridge Local student argumentation and reading strategies. Interna-
Examinations Syndicate and Cambridge University Press. tional Journal of Computer-Supported Collaborative
Schwab, J. (1962). The teaching of science as inquiry. In Learning, 3, 141–164.
J. Schwab & P. Braandwein (Eds.), The teaching of science Wolfe, S., & Flewitt, R. (2010). New technologies, new
(pp. 3–103). Cambridge, MA: Harvard University Press. multimodal literacy practices and young children’s
Sharples, M., Taylor, J., & Vavoula, G. (2007). A theory of metacognitive development. Cambridge Journal of Educa-
learning for the mobile age. In R. Andrews & C. tion, 40, 387–399.
Haythornthwaite (Eds.), Sage handbook of Elearning Yang, K.-L. (2012). Structures of cognitive and
research (pp. 221–247). London: Sage Publications. metacognitive reading strategy use for reading comprehen-
Soller, A. L. (2001). Supporting social interaction in an intel- sion of geometry proof. Educational Studies in Mathemat-
ligent collaborative learning system. International Journal ics, 80, 307–326.
of Artificial Intelligence in Education, 12, 40–62. Yang, X., Yu, S., & Sun, Z. (2013). The effect of collabora-
Song, L., & McNary, S. W. (2011). Understanding students’ tive annotation on Chinese reading level in primary
online interaction: Analysis of discussion board postings. schools in China. British Journal of Educational Technol-
Journal of Interactive Online Learning, 10, 1–14. ogy, 44, 95–111.

© 2014 John Wiley & Sons Ltd

You might also like