Você está na página 1de 26

Computer Assisted Language Learning

Vol. 19, No. 1, February 2006, pp. 79 103

Learner Interaction Management in an


Avatar and Chat-based Virtual World
Mark Peterson*
Tokyo University of Foreign Studies, Japan

In this paper, I report on the findings of a study that investigated non-native speaker interaction in a
three dimensional (3D) virtual world that incorporates avatars and text chat known as Active Worlds.
Analysis of the chat transcripts indicated that the 24 intermediate level EFL participants were able
to undertake a variety of tasks through target language interaction. Moreover, I found evidence in
the corpus that over 3 chat sessions the subjects employed transactional communication strategies
(424) and also interactional strategies (382). Further analysis revealed that instances of negotiation
of meaning focusing on lexis occurred. However, negotiation was infrequent across the 3 task types
administered. Task type appeared to influence the quantity of negotiation, with more negotiation
occurring in the decision-making task than in the jigsaw and opinion-exchange tasks. In addition,
post-study questionnaires indicated that the avatars enhanced the subjects sense of telepresence
and that the learners made use of their communicative features during the interaction. The analysis
further suggested that the use of avatars facilitated learner interaction management during real time
CMC. In this research, I attempted to account for the use of the above strategies and found that the
learner interaction was influenced by the complex interaction of a number of variables including
task type, sociolinguistic factors, context of use and the mix of technical affordances provided by
Active Worlds. This paper concludes by identifying areas of potential for future research in the use of
avatar-based virtual worlds in computer assisted language learning (CALL).

Introduction
In the current study, I investigate the interaction of a group of intermediate level EFL
learners during task-based CMC in Active Worlds. I report on the interaction
management strategies employed, examine how the subjects negotiated meaning and
also attempt to account for the infrequency of negotiation in the data as a whole. In
addition, the discussion focuses on the role of avatars in the interaction. This paper

*Corresponding author. Division of Languages and Information Studies, Tokyo University of


Foreign Studies, 3-11-1 Asahi-cho, Fuchu, Tokyo 183-8534, Japan. Email: mark@tufs.ac.jp
ISSN 0958-8221 (print)/ISSN 1744-3210 (online)/06/01007925
2006 Taylor & Francis
DOI: 10.1080/09588220600804087
80 M. Peterson

concludes with an examination of areas with potential for future research. This
research was motivated by the results of studies on the use of computer-mediated
communication (CMC) tools in CALL. The bulk of findings reported in the current
literature suggest that real time interaction over networks may support aspects of
second language development. Although some recent research has indicated that in
certain contexts, participation in projects involving interaction over networks has no
significant difference on L2 development (Abrams, 2003) researchers have utilized
the recording features of CMC programs and reported that learner interaction in text-
based real time CMC can lead to increased target language output, enhanced
motivation and opportunities for collaborative, learner-centered interaction (Kern,
1995; Kitade, 2000; Warschauer, 1996). These benefits are believed to be the result
of the more equal participation patterns afforded by anonymous online interaction
and the accompanying reduction of inhibition and sociolinguistic cues that can
constrain interaction in face-to-face learning contexts (Hudson & Bruckman, 2002;
Warschauer, Turbee, & Roberts, 1996). Further benefits of learner participation in
text-based real time CMC include opportunities to focus on form (Pellettieri, 2000)
and generate a variety of speech acts (Chun, 1994). Researchers have also attempted
to investigate the characteristics of real time CMC (Murray, 2000; Werry, 1996) and
the communication strategies employed by learners when managing task-based
interaction (Lee, 2002). This effort has been motivated by the need to understand the
ways in which learners avoid and overcome communication problems during tasks, as
studies have indicated that the strategies utilized by learners undertaking this type of
interaction frequently result in the negotiation of meaning (Lee, 2002; Smith, 2003).
Moreover, the literature indicates that there is considerable scope for further research
(Ortega, 1997). Recent studies have explored the role of social and cultural influences
on learner interaction in CMC (Thorne, 2003; Ware, 2005; Warner, 2004) and have
stressed the importance of these factors on language development. One example of a
type of CMC tool that has only recently become the subject of investigation by CALL
researchers is Active Worlds. This virtual reality (VR) tool combines text chat and
avatars (embodied agents) within the framework of a 3D virtual world. The few
learner-based studies have been conducted into the use of Active Worlds (Peterson,
2005; Toyoda & Harrison, 2002) have reported that this example of CMC may
have potential as a learning environment in CALL. This research attempts to build
on this earlier work.1 The following section will examine a distinctive feature of the
Active Worlds tool, avatars.

Background
Avatar-based Virtual Worlds: An Overview
Gerhard, Moore and Hobbs (2004, p. 5) have described an avatar (see Figure 1) as a
user embodiment in a collaborative virtual environment. These user-controlled
embodiments of self first emerged in the 1980s and have since been developed and
incorporated into a number of both text and graphical virtual worlds utilized in
Learner Interaction in a Chat-based Virtual World 81

Figure 1. An Active Worlds guest avatar

education (Schroeder, 2002). Avatars enable a user to adopt a graphical form within a
virtual world, and contribute to the establishment of a sense of presence in a virtual
environment (Svensson, 2003). The sense of feeling present in a virtual world of
being there is known as telepresence (Schroeder, 2002, p. 3). The rationale for
avatars drew on the need to provide users with both a heightened sense of
telepresence and copresence that is a sense of being together (Schroeder, 2002,
p. 4) in a shared virtual environment. Studies suggest that telepresence and
copresence are important influences on online interaction, as they enhance the social,
communication and educational experience provided by collaborative virtual
environments (Ornberg, 2003; Gerhard et al., 2004). An additional motivation
driving development is the fact that these entities provide a means to overcome the
limitations of text-based CMC by providing users with a means to display in real
time, non-verbal communication cues (such as gesture) and emotional states
(happiness) that facilitate communication in face-to-face interaction. A screen
capture of an Active Worlds avatar is reproduced below (Figure 1).
A number of avatar-based virtual environments that incorporate text chat and user-
friendly interfaces have been developed and applied in education. One such
environment (that also encompasses text chat) Active Worlds, has been utilized a
number of small-scale CALL projects.
82 M. Peterson

The Active Worlds Virtual Environment


Active Worlds (www.activeworlds.com) is a browser-based desktop virtual environ-
ment that incorporates a large number of user-created graphical 3D worlds. On
completion of an entry protocol users are transported to the gateway interface; a
central virtual world known as AlphaWorld (see Figure 2, below).
As can be seen in Figure 2, text communication is conducted through a chat
window where messages are posted as they are received by the database in the on-
screen text box (and are also shown in speech bubbles that are displayed by the
senders avatar, this can be seen in Figure 2). As is the case in other forms of real time
text-based CMC messages in the text box are intermixed and strict turn adjacency is
frequently interrupted. Active Worlds provides users with two levels of access.
Registered users can fully customize the appearance of their avatars while guest users
are provided with a standard generic avatar known as a tourist. Although the learners
in this research accessed as tourists, each of them was able to chose an individual
name for their avatar and also select its gender.2 In Active Worlds all individual avatars
regardless of their status, can chat and display in real time aspects of non-verbal
communication such as for example waving. These embodied entities also display
emotional states such as happiness. In addition, they can walk, run and fly while

Figure 2. The Alpha World area of Active Worlds


Learner Interaction in a Chat-based Virtual World 83

traversing virtual space. These behaviors are carried out by means of keyboard
commands or mouse clicks. As a reliable VR world, the researcher considered that
this platform was well suited to the purposes of this project. In the context of CALL,
few studies of learner interaction in avatar-based virtual worlds have been conducted.
The results of these preliminary studies are examined in the following section.

Avatars in CALL: Past Research


A case study conducted by Svensson (2003) reported that native speakers (NSs) and
non-native speakers (NNSs) deployed a variety of speech acts during interaction in
the Alpha World area of the Active Worlds. Svensson further reported (2003, p. 24)
that the avatar-enhanced interaction appeared to engender a high degree of
telepresence among users. Moreover, the subjects communication conformed to a
set of community-based linguistic conventions similar to those found in other forms
of virtual world such as for example MOOs (Cherny, 1999; Weininger & Shield,
2003). These conventions included the use of a highly informal register that
incorporates short utterances, acronyms and the deployment of text to display
physical actions. An experimental study of learner and native speaker (of Japanese)
interaction conducted in Australia by Toyoda and Harrison (2002) reported mixed
results. These researchers reported that in their study that focused primarily on text
chat rather than avatars the subjects had difficulties with the technical aspects of using
Active Worlds. The NNS participants found managing multiple threads with NSs in
real time to be a challenging endeavor (2002, p. 97) due in part to inter-cultural
communication gaps. Moreover, the interruption of turn adjacency (Herring, 1999)
caused by the intermixing of text messages on the screen resulted in a situation where
the subjects made little use of the communication features of their avatars. This
situation was apparently due to the fact that the NNS participants had on some
occasions, difficulties keeping up with the real time interaction. Despite these
problems, the five participants managed their discourse and overcame communica-
tion problems. Although the participants made frequent errors, these mistakes
provided opportunities to engage in the negotiation of meaning in a manner similar in
some respects to that described by Varonis and Gass (1985). A preliminary study of
the text and avatar-based interaction of 15 EFL learners based at a university in Japan
(Peterson, 2005) reported interesting results. In contrast to the findings reported in
Toyoda and Harrison (2002), the subjects in this Active Worlds-based study made use
of the communicative features of their avatars such as waving in order to attract the
attention of potential interlocutors. The use of these communication features
appeared to cease when a dyad was formed and the subjects focused on text chat in
order to undertake the opinion-exchange task. This finding suggests that the avatars
played only a limited role in the task-based interaction. Peterson (2005) further
reported that although a few of the learners had problems with using commands the
transcripts contained evidence of coherent target language discourse. Moreover, the
subjects employed many of the transactional and interactional interaction manage-
ment strategies (see Brown & Yule, 1983) identified in the literature on other
84 M. Peterson

examples of real time CMC that utilize text chat such as MOOs and IRC (Simpson,
2002; Weininger & Shield, 2003; Werry, 1996). For example, the subjects made use
of interactional strategies such as greetings and transactional strategies including
split turns that have been reported in the above studies. Drawing on these results,
I attempted to revisit and expand on the research reported in the above studies.

Research Questions
In this study, I pursued a variety of research questions. I investigated the interaction
of 24 learners during three types of task. Data was collected during three one hour
sessions.
The number of participants and duration of the project were influenced by
institutional constraints.3 During this project a number of additional research
questions emerged. The research questions investigated and their rationales are
described below:

. Research Question 1: What kind of interaction management strategies do NNSs


employ during task-based interaction in Active Worlds? This question was
investigated in order to ascertain if the interaction management strategies
employed were in any way different to those found in previous studies on learner
interaction in Active Worlds (Peterson, 2005; Toyoda & Harrison, 2002) and other
examples of real time CMC (Murray, 2000; Lee, 2002).
. Research Question 2: Do learners make use of the communicative features of their
avatars during task-based interaction in Active Worlds? Previous research had
indicated that learners made little use of the communication features of their
avatars during interaction (Toyoda & Harrison, 2002). This project provided
training in avatar use and I therefore anticipated that the subjects would utilize the
communication features of their avatars during interaction. The above question
was explored through the methodology (outlined in the next section) in order to
determine the ways in which the subjects made use of the communication features
of their avatars. Investigation of this question enabled some tentative assessment
to be made regarding the value of avatars in facilitating interaction management in
Active Worlds.
. Research Question 3: Does the use of a personal avatar enhance the sense of
telepresence in Active Worlds? This question was explored in order to examine the
claim in the literature that the use of avatars enhances the sense of user presence in
virtual worlds and facilitates communication by promoting a sense of involvement
(Svensson, 2003).
. Research Question 4: How do NNSs subjects negotiate meaning when a non-
understanding occurs in Active Worlds? As previous research had established that
negotiation of meaning occurs during interaction in Active Worlds (Toyoda &
Harrison, 2002) and other examples of CMC (Smith, 2003) I anticipated that the
data would show at least limited evidence of negotiation. The above question was
pursued in order to establish if there were any differences between negotiation in
Learner Interaction in a Chat-based Virtual World 85

Active Worlds compared to face-to-face communication and other forms of real


time CMC. Investigating this question offered the additional advantage of
providing insights into the influence of task type on NNS negotiation in
Active Worlds.

Methodology
Participants
The 24 participants in this study (5 males and 19 females) were second, third and
fourth year undergraduates based at a University in Tokyo. The pre-study
questionnaire provided the following information. The subjects ranged in age from
19- to 29-years and the average age was 20-years. The subjects majored in the
following languages, Chinese, Russian, Mongolian, Japanese, Indonesian, Burmese,
Spanish, and French. Four nationalities were present, the majority of the subjects
were Japanese (20) and other nationalities were Chinese (2), Korean (1) and
Malaysian (1). The subjects were intermediate level and this was highlighted in the
recent test scores provided by 20 of the participants. The range of TOEFL scores
provided was between 500 and 540 (paper-based version of TOEFL). These
responses confirmed my observation that the subjects were able to understand
directions and undertake the tasks. The participants reported that they had limited
opportunities to use English outside of the classroom. None of the subjects indicated
they had visited Active Worlds before the project was initiated. The project was
conducted during the fall semester of 2004.

Ethical Issues
All the students who participated in this study were volunteers who took part in the
project in lieu of their lab requirements for the quarter. Before the initiation of this
project, written consent of the subjects was obtained for the collection, analysis and
quotation of the transcript data.

Procedures
This project was conducted in two phases, over a five-week period. In the orientation
phase, the learners undertook (over two one hour sessions) a number of
communication and navigation activities designed to assist familiarization with Active
Worlds. The learners explored the environment, practiced chatting and utilized the
communication features of their avatars such as for example waving. In the second
phase, participants worked in dyads and small groups on a specific task once a week
for 60 minutes.4
The subjects were asked to self-select a partner after they had entered the
environment. On occasion, the composition of dyads and groups was changed due to
absences and other factors. In the sessions, the participants were requested to use
86 M. Peterson

only English. The subjects were located in separate locations in the same computer
room. At the beginning of each session, I provided a handout and explanation of the
class task. During this phase of the project, I observed the interaction and did not
intervene in the subjects online activities.

Tasks
The design of the activities was influenced by the literature on task-based learning.
In particular, the findings of studies which suggest that activities involving
communication can facilitate second language development (see for example Pica,
Kenagy, & Falodun, 1993). The tasks were goal orientated and designed to enable
the subjects to actively engage in purposeful target language interaction. Moreover,
they were designed to provide an authentic context for the interaction. In addition,
the tasks conformed to the curricular goals of the university that stressed the
development of the four skills through communicative teaching methodologies where
English is the primary language of instruction. The participants undertook three task
types. The task types were jigsaw (Session 1), decision-making (Session 2) and
opinion-exchange (Session 3). The tasks were implemented in this sequence in order
to explore contradictions in the literature. Some studies involving learner interaction
in real time CMC (Blake, 2000) have suggested that that jigsaw tasks elicit the
greatest amount of negotiation while other studies (Smith, 2003) have reported that
decision-making tasks yield higher levels. In employing these task types, I attempted
to shed more light on the relationship between task-type and level of negotiation in
real time CMC. In the jigsaw task, six pictures depicting a series of events were mixed
up and divided into two task sheets each containing three pictures. This task required
the subjects uncover the correct story sequence by describing the pictures to each
other. Furthermore, four different low frequency vocabulary items were also included
on each task sheet. The participants were asked to come up with a single valid
solution. In the decision-making task, the subjects were requested to discuss a variety
of possible options in the selection of a gift. Each dyad member was given a task sheet
with four possible presents (low frequency vocabulary items) and was required to
come to a single agreed choice. This task therefore had a variety of possible outcomes.
In the final opinion-exchange task, the learners discussed the qualities of an ideal
marriage partner.

Data Collection, Coding and Analysis


Data collection occurred during the three one-hour chat sessions held in the second
phase of the project. The subjects text chat was logged during each session. Before
the analysis, automatically generated status messages produced by the Active Worlds
system were deleted. In cooperation with the other coder (an experienced instructor
from the same institution) I undertook several readings of the corpus (12, 656
running words) and eleven coding categories in total were identified. These
encompassed discourse management strategies that have been identified in the
Learner Interaction in a Chat-based Virtual World 87

literature on NS and NNS interaction in face-to-face settings and real time CMC.
The following transactional strategies (Brown & Yule, 1983) were identified,
addressivity (Werry, 1996), split turns (Simpson, 2003), time saving devices
(Murray, 2000), feedback markers (Cherny, 1999), clarification and definition
requests (Lee, 2002; Smith, 2003; Varonis & Gass, 1985). Further transactional
strategies (Brown & Yule, 1983) identified included comprehension and confirma-
tion checks (Lee, 2002; Smith, 2003; Varonis & Gass, 1985). In addition, the
following interactional strategies were identified, language routines (Herring, 2001),
use of keyboard symbols to display facial expressions relating to emotional states
(Cherny, 1999) and the use of text to display the effects of intonation (Herring,
2001). In an attempt to achieve inter-coder reliability the following procedures were
undertaken. The two coders (working together) manually coded one third of the data
(the transcripts collected during the first session) following the above framework. The
coders discussed and resolved discrepancies as they arose. The remaining data
(collected during Sessions 2 and 3) was coded individually. In order to facilitate this
stage of the analysis the coders utilised The compleat lexical tutor (http://
132.208.224.131/) online text analysis tool (Sevier, 2004). This tool was considered
appropriate for the purposes of this study as it facilitates the investigation of a corpus
by providing access to a web-based concordancer. The use of this tool enabled each
coder to effectively search the remaining data in order to identify instances of the
strategies under investigation. This process was highly effective as the coders agreed
on 95% of their coding, a finding that indicates a high level of consistency. The final
results of the analysis led to the production of the taxonomy outlined in Tables 1
and 2 of the appendix. This procedure also facilitated the calculation of the total
number of turns during each session, the number of turns involving strategy use and
the number of turns involving negotiation (Table 3). In order to gain further
perspectives on the interaction, I took field notes. In addition, both pre- and post-
study questionnaires were administered.

Results and Discussion


Strategy Use in Active Worlds
In order to answer Research Question 1, (what kind of interaction management
strategies do NNSs employ during task-based interaction in Active Worlds?) I
examined the data collected after each session. Analysis of the transcripts reveled that
the subjects utilized a number of communication strategies in order to manage their
target language interaction. These strategies were of two main types. Transactional
strategies were used to transfer information relevant to the tasks and interactional
strategies were employed in an apparent attempt to establish and maintain social
relationships (Brown & Yule, 1983). As Table 1 shows, a total of eight transactional
strategies were identified. Moreover, transactional strategies were the most frequent
strategy type deployed by the participants (424). The following section will discuss
how these strategies were applied in the context of the negotiation of meaning.
88 M. Peterson

Transactional Strategy Use and Negotiation of Meaning


I found examples of lexical negotiation and these occurred in each of the three task
types. Regarding transactional strategy use, I identified four strategy types that are
associated with the negotiation of meaning (Varonis & Gass, 1985). These strategies
were as follows, clarification and definition requests. Also present were comprehen-
sion and confirmation checks. In this study, negotiations focusing on unknown lexical
items occurred and roughly followed the model of NNS classroom negotiation
(trigger, indicator, response and reaction to response) proposed by Varonis and Gass
(1985). In the majority of these interactions (14 out of a total of 22) non-
understandings were resolved promptly as in the following example drawn from
the opinion-exchange task undertaken in Session 3 that focused on discussing the
qualities of an ideal marriage partner. (Note: All data is reproduced unedited. The
students made all errors. Significant linguistic output is in bold; lines of text not
relevant to the interaction are in parenthesis.)

(1) 1. NNS 1: I dont like the high-handed man


(6 lines of text)
2. NNS 2: high-handed?
(9 lines of text)
3. NNS 1: yeah the person who act without considering others
(10 lines of text)
4. NNS 2: I see. I like sympathetic guy

In the above interaction, the use of the term high-handed by NNS 1 (in line 1) acts as a
trigger for a negotiation event. After six turns of text have scrolled NNS 2 responds with
an utterance (indicator in the above model) designed to display non-understanding.
This statement elicits a response from NNS 1 in line 3. After 10 further lines of text have
scrolled NNS 2 signals (in line 4) that the non-understanding has been resolved. There
were other examples of interaction involving negotiation that differed somewhat from
those reported in Varonis and Gass (1985). These types of negotiations involved
considerable delay between responses. For example, a negotiation that involved a long
delay between turns occurred during the decision-making task implemented during
Session 2. In this task, the subjects were required to share information regarding the
selection of a possible gift (the task sheets incorporated two sets of low frequency
vocabulary items) for a friend from overseas in order to come to a single agreed outcome:

(2) 1. NNS 1: How about wreath!


(7 lines of text)
2. NNS 2: wats wreath?
(31 lines of text)
3. NNS 1: wreath is an arrangement of flowor or leaves
(1 line of text)
4. NNS 2: okay I took it alright
Learner Interaction in a Chat-based Virtual World 89

In line 1, use of the word wreath acts as a trigger for a negotiation. In line 2 NNSs
responds to this trigger by deploying a definition request indicating that a
non-understanding has occurred regarding the meaning of the word wreath. After
a long delay (31 lines of text), NNS 1 replies with an utterance containing a repetition
of the problematic word followed by an explanation (reaction to a response in the
above model). This utterance appears to resolve the situation, as in line 4 NNS 2
promptly replies with an utterance indicating that understanding of the previously
unknown item has been achieved. The above interaction highlights the fact that long
delays occurred between the indicator and response during eight interactions
involving negotiation. There were also examples of negotiation in which there was a
long delay between the trigger and indicator a phenomenon reported in a study of
learner real time text-based interaction conducted by Smith (2003). For example, in
Session 1 during the jigsaw (picture description task) task the following series of
interactions (involving three learners) occurred:

(3) 1. NNS 1: no. 1 priorty?


(31 lines of text)
2. NNS 2: by the way, what does NO1 PRTORITY MEAN?
(33 lines of text)
3. NNS 3: I think it means the most important policy, the most
important point of education
(12 lines of text)
4. NNS 2: thank you, now I see
(13 lines of text)
5. NNS 1: thats right

In the above interaction, the utterance made by NNS 1 acts a trigger for a
negotiation. However, it takes NNS 2 a considerable period of time (31 turns) to
respond with a question (indicator) that signals that a non-understanding has
occurred. In an example of transactional strategy use, NNS 2 deploys a definition
request (in line 2) that contains upper case text to display emphasis, a strategy that
has been reported in studies of learner interaction in text-based real time CMC
(Murray, 2000; Simpson, 2002). In a further interesting aspect of this interaction,
after 33 lines of text have scrolled the response does not come first from NNS 1 (who
may have been confused by the spelling error made by NNS 2 in line 2) but from
another learner (NNS 3) and this utterance appears to resolve the matter as the
reaction (to a response) provided in line 4 signals that understanding has been
achieved. Examples two and three indicate that long delays can occur during
negotiation of meaning in Active Worlds. Despite the fact that there were occasions in
which there were considerable delays between messages, the learners in this study
appeared for the most part able to keep up with the interaction and provide an
appropriate response. Moreover, once a dyad had been formed there was little
evidence of the topic decay that has been reported in some studies of text-based real
time CMC (Herring, 1999). During the interaction, the subjects managed to
90 M. Peterson

produce coherent target language discourse relevant to the tasks. A partial


explanation for this situation may be the presence of personalized mobile avatars
in Active Worlds. This aspect of the interaction will be examined at a later stage of the
discussion.
The findings of this study also highlight the potential influence of task type on
lexical negotiation in Active Worlds. As can be seen in Table 3, the decision-making
task yielded a higher number of negotiations (12 from a total of 22) than the other
task types, a result similar to that reported by Smith (2003). This is a somewhat
surprising finding as the literature suggests that in face-to-face contexts, jigsaw tasks
elicit a higher degree of negotiation than other task types (Pica et al., 1993). A possible
cause may be that as in Smiths study, the decision-making task required information
exchange involving discussion of low frequency vocabulary items. Such activity was
essential for task completion. This situation therefore led the subjects to focus on
these items. Although the jigsaw task also required information exchange discussion
of the vocabulary items was not essential to task completion. This may have resulted
in a situation where the subjects viewed completing the task as an imperative and led
them to view discussion of the vocabulary items as a less important activity (Smith,
2003, p. 45).
The examples discussed in this section demonstrate how the learners negotiated
meaning when a non-understanding occurred (Research Question 4). However, as
may be seen in Table 3, strategies associated with negotiation accounted for only 22
of the 424 transactional strategies identified. Moreover, as Table 3 demonstrates,
negotiation was infrequent in the data as a whole. The low incidence of negotiation in
this study contrasts with the findings reported by Toyoda and Harrison (2002), who
found 45 instances. This variance in results can be attributed, in part, to differences
in project configurations and learner proficiency levels. Restrictions on lab access
made it impossible to schedule this project for longer than five sessions. In contrast,
the project conducted by the above researchers was conducted over a semester rather
than a quarter and therefore provided greater opportunities for the subjects to
negotiate meaning. Furthermore, this study (unlike the above research) did not utilize
native speaker participants, as they were unavailable. Some studies have reported that
the presence of NS interlocutors in CMC-based CALL projects involving NNSs
provides scaffolding and results in high levels of negotiation (see for example
Lee, 2002). In addition to the above factors, the infrequency of negotiation in the data
may have been due in part, to the use of avoidance strategies that were the result of
the subjects reluctance to indicate non-understanding. This behavior could have
been motivated by a desire to maintain status with peers (see the discussion on
politeness). This partial explanation for the absence of negotiation in much of the
data is plausible, given that avoidance behaviors have been reported in some studies
of intermediate level NNS NNS interaction in face-to-face communication
(Meierkord, 2000). An additional factor that may have contributed to the infrequency
of negotiation was the online nature of the interaction. As the majority of the in-
teraction occurred in real time via typed text, in situations when messages were
scrolling rapidly there remains the possibility that there were instances where the
Learner Interaction in a Chat-based Virtual World 91

subjects ignored utterances they did not understand due to a need to keep up with the
discourse (Lee, 2002, p. 284).

Transactional Strategy Use: Addressivity, Time Saving Devices,


Split Turns and Feedback Markers
As Table 1 shows, the subjects deployed a series of adaptive transactional strategies
that facilitated information exchange under conditions in which messages are
intermixed and scroll in real time. The learners made use of addressivity, that is, the
strategy of explicitly naming an intended message recipient (Werry, 1996, p. 52).
Addressivity was utilized as a turn tracking device that enabled the participants
to quickly obtain task partners at the beginning of a session (Example 4 from
Session 3):

(4) 1. NNS 1: hi, hana


(3 lines of text)
2. NNS 2: Hi, Elif!
(3 lines of text)
3. NNS 1: I have 2
(9 lines of text)
4. NNS 1: lets talk together
(3 lines of text)
5. NNS 1: ok, Elif!

and was further deployed to efficiently exchange information relevant to the tasks
(and maintain contact) over extended periods of interaction (Example 5 from
Session 2):

(5) 1. NNS 1: and what is the last thing? samurai princess


(2 lines of text)
2. NNS 2: okay maria, what didnt you get?
(3 lines of text)
3. NNS 2: Magnifying glass
(4 lines of text)
4. NNS 2: do you know what it is? maria?
(4 lines of text)
5. NNS 1: besides the extension cord and the comb and the bouquet
(24 lines of text)
6. NNS 1: samurai princess?
(7 lines of text)
7. NNS 2: its for the elder ppl, when they read newspapers and stuff, they use
this glass
8. NNS 2: maria?
(10 lines of text)
92 M. Peterson

9. NNS 1: I got it!


(4 lines of text)
10. NNS 2: okay!

The above interaction also displays a further example of transactional strategy use the
employment of split turns (lines 7 and 8). These were designed to supply information
and move the discourse forward (16 instances were recorded see Table 1). The most
frequent transactional strategy employed by the subjects, accounting for 201 from a
total of 424 strategies (see Table 1) was the use of discourse markers in what was
likely an effort to provide transactional feedback. These markers that have been
identified in other studies of real time CMC (Werry, 1996) included interjections
(ok, yes, wow) and other forms of utterance (um, oh, erm). These markers appeared
designed to signal the attention state of an interlocutor and also displayed awareness
that the interaction was ongoing (Cherny, 1999). The data further contained time
saving linguistic devices (80 instances were recorded see Table 1) that were a product
of the need to respond promptly in instances when messages were scrolling rapidly.
Time saving devices included abbreviations such as u (for you) and r (for are),
contractions (see line 4 in Example 4) and acronyms including the well-known
lol (Murray, 2000).
The use of addressivity contributed to the maintenance of coherent interaction
for the duration of each session. As can be observed in Tables 1 and 2 addressivity was
the third most frequent strategy deployed by the subjects accounting for 24.8% of
total transactional strategies. However, as Table 1 shows addressivity was employed in
only 105 occasions over the 2002 turns that were recorded over the three sessions
(see Table 3). Examination of the data and the researchers field notes confirmed that
were periods during the interaction in which the subjects made little use of addres-
sivity. There are a number of possible explanations for the low incidence of this
strategy. One possibility was that as the database automatically places the name of a
user before their message some of the learners were tracking these names as they
scrolled in the text box. A further possibility was that the subjects were moving their
avatars as a means to facilitate turn tracking. I observed that the subjects quickly
moved away from the central area of Alpha World after they arrived, as messages in the
speech bubbles were difficult to read when many avatars congregated. This obser-
vation was confirmed in several (4) responses to the post-study questionnaire were
subjects claimed that messages were hard to follow when many avatars congregated in
a single area. In an attempt to make communication easier the learners (after they had
found a task partner) moved their avatars to different locations in Alpha World. Moving
to another area made it easier to read the speech bubbles produced by an interlocutors
avatar and this behavior may have reduced the frequency of addressivity. The ability of
individual users to move their avatars also accounts (in part) for the phenomenon
reported earliernamely the fact that the participants were able to produce coherent
target language output focused on the tasks. In addition to the above strategies, the
subjects utilized a number of interactional management strategies that highlight
possible technical and sociolinguistic influences on the interaction.
Learner Interaction in a Chat-based Virtual World 93

Interactional Strategy Use


Interactional communication strategies are a well-documented feature of face-to-face
interaction (Brown & Yule, 1983). Researchers have also identified the presence of
these strategies in text-based NS network interaction (Rintel & Pittam, 1997). The
data collected in this study contained evidence for interactional communication
strategies.
The following sections will describe these strategies and discuss possible
motivations for their use.

Interactional Strategy Use: Politeness


As Table 2 shows, three types of interactional strategy were identified in the data, the
most frequent of these being the use of language routines (Rintel, Mulholland, &
Pittam, 2001), incorporating politeness strategies such as greetings (Example 6) and
leave takings (Example 7):

(6) 1. NNS 1: hi hana


(3 lines of text)
2. NNS 2: Hi, Elif!
(13 lines of text)
3. NNS 1: lets talk together
(3 lines of text)
4. NNS 2: ok, Elif!

(7) 1. NNS 1: nice chatting with u jaca


(14 lines of text)
2. NNS 2: see you

During the interaction, the subjects employed both positive and negative politeness
strategies (Brown & Levinson, 1987). These strategies may have been implemen-
ted in order to establish and maintain a positive face to interlocutors (Goffman,
1959) and also as a means to reduce the risks (to face) of potentially threatening
speech acts (such as for example requesting information) that were a requirement
of the tasks. Positive politeness behaviors included strategies designed to
emphasize a common identify and in-group status. I found instances of these
strategies in the data including the use of humor (Example 8), colloquial
expressions (Examples 9 and 10), small talk (Example 11), and inclusive forms
such as we (Example 12):

(8) 1. NNS 1: Does he have a monkey face?


(11 lines of text)
2. NNS 2: mmm.maybe . . . doesnt he?
(11 lines of text)
94 M. Peterson

3. NNS 1: my friend said he is like a gorilla


(9 lines of text)
4. NNS 2: oh, thats right!! HAHAHA!!!

(9) NNS: didnt you say that we are gonna get corkboard for him, maria?

(10) NNS: i wanna get a partner who is not japanese

(11) 1. NNS 1: Pam!!


(3 lines of text)
2. NNS 2: rin!
(7 lines of text)
3. NNS 1: Long time no see!!
4. NNS 2: im glad to see U again!

(12) NNS: we need more opportunities to excericse!

In addition, negative politeness behaviors were identified. In face-to-face commu-


nication, these are strategies designed to signal social distance, respect for the
addressee and the desire to avoid imposition. In this study, negative politeness
behaviors included the use of formal language (13) and apologies (14):

(13) 1. NNS 1: so, shall we proceed to the next step?


(5 lines of text)
2. NNS2: yes! Step 3!!

(14) 1. NNS 1: did you get the presents of mine_?


(10 lines of text)
2. NNS 2: no . . .
(9 lines of text)
3. NNS 1: sorry then ill tell it to you again right?
(4 lines of text)
4. NNS 2: thank you!!

Table 2 demonstrates that politeness behaviors accounted for 277 of the 382
interactional strategies identified in the corpus. One possible explanation for the
prevalence of these strategies lies in the text-based nature of a significant quantity of
the interaction, where the subjects had to guard against the possibility of their
messages being misunderstood in the absence of intonation and many of the non-
verbal cues that influence interaction in face-to-face communication (Ware, 2005).
A further explanation for the frequent use of politeness can be found in sociocultural
factors. The majority of the subjects (20 out of 24) were Japanese, a culture where
maintaining status with peers is an important influence on communication (Lebra,
1976; Matsumoto, 1988). This fact, combined with the L2 competence level of the
Learner Interaction in a Chat-based Virtual World 95

subjects could have led to the transfer of L1 practices such as the use of politeness to
the online medium. Moreover, as this project was undertaken in Japan the
interactional context, an important influence on interaction in CMC (Baym, 1995)
may have further contributed to the frequent utilization of politeness strategies.
A final partial explanation for the application of these strategies could lie in the fact
that the subjects were aware that they were being observed and that the interaction
was being recorded.

Interactional Strategy Use: Deployment of Keyboard Symbols


The subjects deployed interactional interaction management strategies that were
adapted to the particular communication context provided by Active Worlds.
Table 2 shows that the use of keyboard symbols to display facial expressions and
emotional states occurred on 67 occasions. As the avatars cannot display facial
expressions or use intonation and have the capacity to show only a limited
repertoire of emotions (such as joy) and physical actions (waving) the subjects
employed a variety of text and keyboard symbols in an attempt to display these
aspects of communication. The questionnaire data revealed that the subjects
claimed to have made use of the above features of their avatars during communi-
cation. This finding was corroborated by my observations and lends support to the
view that during periods of the interaction the participants utilized the above
features of their avatars. However, the learners further reported that they also made
use of keyboard symbols in order to display the above aspects of communication.
The participants used combinations of keyboard symbols (smileys) to display
emotional states (happiness) and facial expressions (surprise) as may be observed in
the following examples:

(15) NNS: really? ^o^

(16) NNS: yeah . . . people say that the looks is not so important compared with the
character, but i think as well that we need looks ^_^

(17) NNS: thanks!!(^-^)///

There were also examples of the smileys that are typical of Japanese NS usage in
CMC. Examples of these can be seen in the following interactions where they assisted
in the completion of the tasks by providing positive feedback:

(18) 1. NNS 1: billy should be . . . given nagnigying glass or razor you have^^ oh i
see^_^ so do we have to chose only billys present?
(16 lines of text)
2. NNS 2: ok!

(19) NNS: should we talk about the part2 . . . ^_^;


96 M. Peterson

In a further example of interactional strategy use on 38 occasions (see Table 2) the


participants utilized multiple text characters to display the effects of intonation
(Simpson, 2002):

(20) NNS: your eyes are sooooooooooo scary!!!!!!!!!!

In addition, the participants deployed upper case (in combination with exclamation
marks) for emphasis:

(21) 1. NNS 1: I like boys who can make a joke


(9 lines of text)
2. NNS 2: YES!! I strong agree with you!Elif

and to gain attention:

(22) NNS: HELP ME !!!!!!!!!!!!!!1

(23) NNS: DONT IGNORE ME!!!!!!!!!!!!!1

In a further possible explanation for the use of keyboard symbols as an interactional


communication strategy, responses to the pre-study questionnaire indicated that all of
the subjects had used various types of Internet chat tool before the initiation of the
project. This fact raises the possibility that the inability of the avatars to display facial
expressions led to situations where the subjects transferred communication strategies
they had already used in other examples of CMC to the Active Worlds environment.

Limitations of this Research


There are a number of issues that must be considered when seeking to generalize the
results of this study. Institutional constraints including restrictions on lab access and
the number of participants, combined with the fact that the sessions could only be
scheduled once a week limited the duration of the project. The fact that the task phase
of this project could only be held over three sessions highlights the need to exercise a
degree of caution in interpreting the results. In addition, an absence of funding
meant that it was not possible to utilize the building features of Active Worlds. This
was a limitation, given the fact that several studies highlight the beneficial nature of
learners L2 interaction during building activities in virtual worlds (see for example
Von Der Emde, Schneider, & Kotter, 2001). A further potential limitation may be the
fact that aspects of the data display the influence of training effects. For example, it
was observed that the subjects made use of the communication features of their
avatars to a greater degree than was the case in the studies conducted by Peterson
(2005) and Toyoda and Harrison (2002). Learner feedback suggests that the
presence of avatars facilitated the communication however, when interpreting (and
generalizing) data collected in virtual worlds there is a need to acknowledge that
Learner Interaction in a Chat-based Virtual World 97

learner behaviors can be significantly effected by the complex interplay of factors that
influence interaction in real time CMC. The literature on learner interaction in real
time CMC suggests that results of individual studies can vary greatly depending on
variables such as for example affective factors, project configuration, technical
features of the CMC tool employed and the particular institutional context where a
study is conducted (Hudson & Bruckman, 2002; Thorne, 2004; Ware, 2005).

Conclusions
The investigation of the data was influenced by the results reported in literature on
the application of Active Worlds and other examples of real time CMC in CALL
projects involving second language learners. A number of findings emerged that shed
new light on the research questions that were the focus of this study.

. Research Question 1: What kind of interaction management strategies do NNSs


employ during task-based interaction in Active Worlds?

As can be seen in Tables 1 and 2 the subjects employed a number of strategies that
have been identified in the literature of learner interaction in face-to-face
communication and real time text-based CMC. Analysis of the data demonstrates
that the subjects employed adaptive transactional management strategies including
the use of feedback markers (201), addressivity (105) and time saving devices (80).
These strategies were deployed in an attempt to exchange information relevant to the
tasks. In a significant finding, addressivity was deployed in only 5% of the total turns
(see Tables 1 and 3). This result may be attributable to design of the environment. As
was noted in Section 6.3, the Active Worlds database places a users name before their
utterance after it has been sent as means to facilitate turn tracking. However, there
remains the possibility that the presence of personal mobile avatars reduced the need
for addressivity as the subjects could move to less crowded areas and this made the
comments produced in the speech bubbles easier to read. The data further revealed
the presence of interactional strategies the most frequent of which was the use of
language routines incorporating positive and negative politeness behaviors (277). The
motivations for the use of these strategies are difficult to determine with certainty. My
observations of the interaction and analysis of the transcripts coupled with learner
feedback indicate a number of possible motivations for their deployment. Explana-
tions include the text and avatar-based nature of the communication in Active Worlds
(where intonation and most paralinguistic features were absent or reduced) led to the
transfer of L1 face-to-face communication practices such as politeness to the online
medium. Sociolinguistic factors such as a concern to maintain face (an important
influence on communication in Japanese culture) and a desire to avoid misunder-
standings may also partially account for the use of these strategies. There remains the
further possibility that the institutional context of this study influenced the subjects
strategy use given the fact that they were aware that their interaction was being
recorded and observed. In a positive finding, the application of these strategies
98 M. Peterson

appeared to assist the learners in the production of coherent target language output
(that incorporated adjacency pairs) focused on the tasks.

. Research Question 2: Do learners make use of the communicative features of


their avatars during task-based interaction in Active Worlds?

As can be observed in Table 3, less than half of the turns analyzed contained evidence
of communication strategy use. This finding indicates that during periods of the
interaction a number of the subjects made use of the communication features of their
avatars. Responses to the post-study questionnaire indicated that 15 of the subjects
made use of the communication features of their avatars. The majority of subjects (15)
employed waving in order to attract the attention of potential task partners. Of these
subjects, 13 used the features of their avatar that enabled emotional responses such as
for example joy to be displayed. However, nine of the subjects claimed that they did not
make any use of the above communication features. A variety of reasons were provided
for these negative responses. Six of the subjects claimed that they had on occasion,
difficulties keeping up with the interaction due to the rapid scrolling of messages,
limited typing skills and gaps in vocabulary knowledge. This resulted in a situation
where they had no time. Two subjects further reported that the communication
features of their avatars were too limited to be of value during the interaction. The
above responses may therefore account for the presence of feedback symbols in the data
(see Table 1). In contrast to these negative responses, fourteen of the subjects claimed
to have continued to use the communication features of their avatars during
communication. In a further interesting result, nine of the learners claimed that being
able to move their avatar was useful as they could move from crowded areas and this
made it easier to follow their interlocutors responses. This finding suggests that the
ability of individual avatars to move may facilitate turn tracking.

. Research Question 3: Does the use of avatars enhance the sense of telepresence
in Active Worlds?

In order to investigate the above research question, the subjects were asked in the
pre-study questionnaire to state if they had experienced a sense of telepresence and
copresence when using other chat environments such as MOOs and IRC. In their
responses, 11 of the subjects claimed to have experienced a sense of telepresence
when using other chat environments while 10 subjects reported a sense of copresence.
In the post-study questionnaire, the subjects were asked if using a personal avatar in
Active Worlds made them feel a sense of telepresence. Seventeen learners responded
to this question positively. In their comments these learners stated that they
experienced an enhanced sense of telepresence when using their avatars. Fifteen of
the subjects claimed that when utilizing their avatars they experienced a high degree
of copresence with their interlocutors. Three of the subjects claimed that having an
avatar did not make them feel more present in the environment, while a further two
subjects gave ambiguous responses. These findings suggest that for a majority of the
Learner Interaction in a Chat-based Virtual World 99

participants, the use of a personal avatar contributed to the creation of a sense of


telepresence and copresence within Active Worlds. Moreover, the use of an avatar
appeared to enhance these aspects of the interaction to a greater degree than was
reported for other chat environments.
In an additional finding, five of the subjects observed that the use of avatars made
them feel more involved in the interaction. These subjects further noted that
interacting using avatars in Active Worlds was an interesting experience. Finally, 14 of
the participants claimed that using avatars to communicate was an enjoyable
experience. These subjects also indicated that they wanted to take part in future
projects involving avatar-based virtual worlds.

. Research Question 4: How do NNSs subjects negotiate meaning when a


non-understanding occurs in Active Worlds?

In a positive finding, analysis of chat the transcripts (see Table 3) showed that the
subjects engaged in the negotiation of meaning during each of the task sessions.
Interactions involving negotiation of new lexis occurred in a manner similar to that
reported in studies of face-to-face interaction (Varonis & Gass, 1985) and other
examples of real time CMC (Lee, 2002). During these interactions the learners
employed confirmation and comprehension checks. They also utilized definition and
clarification requests. As was reported in Smith (2003) on occasion there were long
delays between turns involving triggers and indicators. Moreover, there were also
delays between indicators and responses (see examples in section 6.2). However,
these delays did not appear to prevent negotiation occurring. The results also
highlight the possible relationship between task type and quantity of negotiation, as
more negotiations were recorded during the decision-making task than in the other
task types. This finding is consistent with the results of other studies involving learner
interaction in real time CMC (Smith, 2003). A further significant aspect of the
interaction was the low incidence of negotiation in the corpus as a whole. Although
the reasons for this result cannot be determined with certainty, there remains a
possibility that avoidance strategies coupled to difficulties keeping up with the
interaction (on occasions when messages were scrolling rapidly) contributed to
this finding.

Future Directions
Although this study was subject to the limitations described in Section 7, the results
raise a number of issues that may be the subject of investigation in future studies. As
negotiation was limited, research efforts could analyze the interaction of more diverse
learner groups in projects of longer duration (and that utilize a wider variety of tasks),
in order to investigate if variations in project configuration result in the more
extensive application of interaction management strategies associated with negotia-
tion. The results of this research further draw attention to the importance of
appreciating the influence of learning context in CALL projects involving real time
100 M. Peterson

online interaction. The use of the communication features of avatars is an area that
also invites examination. Although it was not possible in this project, filming the
online avatar-based interaction of learners may provide additional insights into the
role these entities play in real time interaction management in virtual worlds.
Moreover, the presence in the data of symbols that were employed to display
emotional states and facial expressions highlights the limitations of the avatars utilized
in Active Worlds. Recent advances in avatar technologies in which paralinguistic cues
such as facial expressions, head and eye movements are displayed in real time present
researchers with new opportunities to explore the influence of these aspects of
interaction management in virtual worlds. In conclusion, this paper highlights the
complex nature of learner interaction management in Active Worlds and highlights the
need for further studies that explore the factors that influence second language
interaction and development in online virtual worlds.

Notes
1. As has been noted by Levy (1997, p. xi) the frequent failure to follow up on the results of past
studies remains a major obstacle to advances in CALL research.
2. Due to a lack of funding it was not possible to provide the subjects with fully personalized
avatars.
3. Restrictions on lab access meant that the project could only be scheduled over five weeks.
4. At the university where this research was conducted and as in most Japanese university language
programs, classes are scheduled only once a week.

Notes on contributor
Mark Peterson is a Senior Lecturer in the Department of Languages and Information
Studies, Tokyo University of Foreign Studies.

References
Abrams, Z. (2003). The effect of synchronous and asynchronous CMC on oral performance in
German. The Modern Language Journal, 87, 157 167.
Baym, N. (1995). The emergence of community in computer-mediated communication. In
S. Jones (Ed.), Cybersociety: computermediated communication and community (pp. 138 163).
Thousand Oaks, CA: Sage.
Blake, R. (2000). Computer mediated communication: a window on L2 Spanish interlanguage.
Language Learning & Technology, 4(1), 120 136. Retrieved May 1, 2005, from http://
lit.msu.edu/vol4num1/blake/default.html
Brown, P., & Levinson, S. (1987). Politeness: some universals in language usage. Cambridge:
Cambridge University Press.
Brown, G., & Yule, G. (1983). Discourse analysis. Cambridge: Cambridge University Press.
Cherny, L. (1999). Conversation and community: chat in a virtual world. Stanford: CSLI Publications.
Chun, D. (1994). Using computer networks to facilitate the acquisition of interactive competence.
System, 22(1), 17 31.
Gerhard, M., Moore, D., & Hobbs, D. (2004). Embodiment and copresence in collaborative
interfaces. Human-Computer Studies, 61(4), 453 480.
Goffman, E. (1959). The presentation of self in everyday life. NY: Doubleday.
Learner Interaction in a Chat-based Virtual World 101

Herring, S. C. (1999). Interactional coherence in CMC. Journal of Computer Mediated


Communication, 4(4). Retrieved October 22, 2005, from http://www.ascusc.org/jcmc/vol4/
issue4/herring.html
Herring, S. C. (2001). Computer-mediated discourse. In D. Schiffin, D. Tannen, & H. Hamilton
(Eds.), The handbook of discourse analysis (pp. 612 634). Oxford: Blackwell.
Hudson, J. M., & Bruckman, A. S. (2002). IRC Francais: the creation of an Internet-based SLA
community. Computer Assisted Language Learning, 15(2), 109 134.
Kern, R. G. (1995). Restructuring classroom interaction with networked computers: effects on
quantity and characteristics of language production. The Modern Language Journal, 25(5),
441 454.
Kitade, K. (2000). L2 learners discourse and SLA theories in CMC: collaborative interaction in
Internet chat. Computer Assisted Language Learning, 13(2), 143 166.
Lebra, T. S. (1976). Japanese patterns of behavior. Honolulu, HI: University Press of Hawaii.
Lee, L. (2002). Synchronous online exchanges: a study of modification devices on non-native
discourse. System, 30(3), 275 288.
Levy, M. (1997). Computer assisted language learning: context and conceptualization. Oxford:
Clarendon Press.
Matsumoto, Y. (1988). Reexamination of the universality of face. Journal of Pragmatics, 12, 403 426.
Meierkord, K. (2000). Interpreting successful lingua-franca interaction. An analysis of non-native/
non-native small talk conversation in English. Linguistik, 5, 1 11. Retrieved October 16,
2005, from http://www.linguistik online.de/1_00/index.html
Murray, D. E. (2000). Protean communication: the language of computermediated communica-
tion. TESOL Quarterly, 34(3), 397 421.
Ornberg, T. (2003). Linguistic presence on the Internet: communication, worldview and presence in online
virtual environments. Unpublished masters thesis, University of Umea.
Ortega, L. (1997). Processes and outcomes in networked classroom interaction: defining the
research agenda for L2 computer-assisted classroom discussion. Language Learning &
Technology, 1(1), 82 93. Retrieved October 22, 2005, from http://llt.msu.edu/vol1num1/
ortega/default.html
Pellettieri, J. (2000). Negotiation in cyberspace: the role of chatting in the development of
grammatical competence. In R. Kern, & M. Warschauer (Eds.), Network-based language
teaching: concepts and practice (pp. 59 86). Cambridge: Cambridge University Press.
Peterson, M. (2005). Learning interaction in an avatar-based virtual environment: a preliminary
study. PacCALL Journal, 1(1), 29 40.
Pica, T., Kanagy, R., & Falodun, J. (1993). Choosing and using communication tasks for second
language instruction. In G. Crookes & S. Gass (Eds.), Tasks and language learning: integrating
theory and practice (pp. 9 34). Clevedon: Multilingual Matters.
Rintel, E. S., Mulholland, J., & Pittam, J. (2001). First things first: Internet relay chat openings.
Journal of Computer-Mediated Communication, 6(1). Retrieved March 2, 2005, from http://
www.ascusc.org/jcmc/vol6/issue3/rintel.html
Rintel, E. S., & Pittam, J. (1997). Strangers in a strange land: interaction management on Internet
Relay Chat. Human Communication Research, 23, 507 534.
Schroeder, R. (2002). Social interaction in virtual environments: key issues, common themes, and a
framework for research. In R. Schroeder (Ed.), The social life of avatars: presence and interaction
in shared virtual environments (pp. 1 16). London: Springer.
Sevier, M. (2004). The compleat lexical tutor, v 4. TESL EJ, 8 (3). Retrieved June 17, 2005, from:
http://www writing.berkeley.edu/TESL-EJ/ej31/m2.html
Simpson, J. (2002). Discourse and synchronous computer-mediated communication: uniting
speaking and writing. In P. Thompson, & K. Spelman Miller (Eds.), Unity and diversity in
applied linguistics. Studies in applied linguistics 17. London: BALL and Continuum.
Smith, B. (2003). Computer-mediated negotiated interaction: an expanded model. The Modern
Language Journal, 87, 38 57.
102 M. Peterson

Svensson, P. (2003). Virtual worlds as arenas for language learning. In U. Felix (Ed.), Language
learning online: towards best practice (pp. 123 142). Lisse: Swets & Zeitlinger.
Throne, S. (2004). Artifacts and cultures-of-use in intercultural communication. Language Learning
and Technology, 7(2), 38 67. Retrieved August 20, 2005, from http://llt.msu.edu/vol7num2/
thorne/default.html
Toyoda, E., & Harrison, R. (2002). Categorization of text chat communication between learners
and native speakers of Japanese. Language Learning and Technology, 6(1), 82 99. Retrieved
August 22, 2005, from http://llt.msu.edu/vol6num1/toyoda/
Varonis, E., & Gass, S. (1985). Non-native/Non-native conversations: a model for negotiating
meaning. Applied Linguistics, 6(1), 71 91.
Von Der Emde, S., Schneider, J., & Kotter, M. (2001). Technically speaking: transforming
language learning through virtual learning environments (MOOs). The Modern Language
Journal, 85, 211 225.
Ware, P. (2005). Missed communication in online communication: tensions in a German
American telecommunication. Language Learning and Technology, 9(2), 64 89. Retrieved
September 2, 2005, from http://llt.msu.edu/vol9num2/ware/default.html
Warner, C. N. (2004). Its just a game, right? Types of play in foreign language CMC. Language
Learning and Technology, 8(2), 69 87. Retrieved August 22, 2005, from http://llt.msu.edu/
vol8num2/warner/default.html
Warschauer, M. (1996). Motivational aspects of using computers for writing and communication.
In M. Warschauer (Ed.), Telecollaboration in foreign language learning (pp. 29 44). Hawaii:
University of Hawaii Press.
Warschauer, M., Turbee, L., & Roberts, B. (1996). Computer learning networks and student
empowerment. System, 24(1), 1 14.
Weininger, M., & Shield, L. (2003). Promoting oral production in a written channel: an investi-
gation of learner language in MOO. Computer Assisted Language Learning, 16(4), 329 349.
Werry, C. (1996). Linguistic and interactional features of Internet relay chat. In S. C. Herring
(Ed.), Computer-mediated communication: linguistic, social and cross-cultural perspectives
(pp. 47 63). Amsterdam: John Benjamins.

Appendix

Table 1. Transactional strategy use over three task sessions

N 22 Percentage of all
Strategy (all students) transactional strategies

Clarification requests 6 1.4%


Comprehension checks 4 0.9%
Confirmation checks 8 1.9%
Definition requests 4 0.9%
Split turns 16 3.8%
Feedback markers of attention 201 47.4%
and attitude
Time saving devices 80 18.9%
Addressivity 105 24.8%
Total 424 100
Learner Interaction in a Chat-based Virtual World 103

Table 2. Interactional strategy use over three task sessions

N 22 Percentage of all
Strategy (all students) interactional strategies

Language routines incorporating positive 277 72.5%


and negative politeness
Use of keyboard symbols to display facial 67 17.5%
expressions and emotional states
Use of text to display the effects 38 10.0%
of intonation
Total 382 100

Table 3. Total number of turns, turns involving strategy use and negotiation

Total turns Total turns


involving involving
use of the strategies Percentage
Total strategies associated of all
number identified with turns
of turns (all in tables 1 negotiation involving
Session Task type subjects) and 2 negotiation

1 Jigsaw 596 258 6 1.0%


2 Decision-making 717 279 12 1.7%
3 Opinion-exchange 689 269 4 0.6%
Total 2002 806 22

Você também pode gostar