Team Now & Later Presentation

Check out our presentation. Please comment. Thanks!

Team Now & Later Presentation

View more presentations from Julius.

Project Proposal: The Language of Emerging Leadership

Advances in computer-mediated communication tools have made working in distributed teams easier than ever. We can use email to transfer documents in progress, video or audio conferencing to have weekly meetings, and instant messaging for quick questions and answers. However, as we introduce more layers of technology between group members, the social dynamics of group work change. One such social dynamic is leadership. It is well established in the literature that leaders play a critical role in group work and that the presence of an effective leader increases group productivity, but technology mediation as well as spatial distance introduce additional challenges for leaders of distributed teams over leaders of collocated teams [1,2].

Leaders can be assigned or they can emerge as group work progresses, but regardless group members should be able to identify them as leaders if they are to be effective in their leadership role [5]. In face-to-face settings, leaders make their presence known in a number of ways, including their seat location in meetings, body language and voice inflection [7]. However, in computer-mediated communication, these cues are either limited or non-existent. In text-based communication mediums that are often used to coordinate group work such as e-mail, instant messaging, wikis, and forums, leaders only have language available to them as a way of making their presence known. In addition to establishing their presence, an important role of a leader is to establish group trust and cohesion [6]. While higher levels of trust and cohesiveness amongst team members are also associated with higher productivity, it is challenging to develop these dimensions without the aid of the rich communication that face-to-face interaction provides [6]. This is yet another challenge for leaders of virtual teams, particularly for those that communicate primarily (or solely) using text-based mediums.

We are interested in studying the ways that leaders emerge in these text-based environments and how this compares to the ways that assigned leaders lead. Following the Social Language Processing (SLP) framework [4] we plan on collecting and analyzing transcripts of groups completing tasks via a chat based interface to identify linguistic features that can predict the social dynamic of leadership. While it has been suggested that leaders of virtual teams should receive special training in order to effectively lead teams utilizing computer-mediated communication tools [6,7], we believe that the automatic identification of emergent and assigned leaders would allow us to provide tailored tools facilitating leadership in these groups.

To collect data, we will recruit 80 participants to take part in our study. Participants will work in groups of four (20 groups in total) using a chat interface to complete a task that involves coming to a consensus. We are still considering the exact task, however we find the murder mystery task as used in [2] promising because it involves the coordination of information to reach a consensus and it does not require the manipulation of physical objects as does solving a puzzle (we do not want to add the additional dimension of visual information into our study at this time). We will have two conditions; in one condition a leader will be assigned and the group will be told that this person is the leader and in the other condition a leader will be assigned, but their group will not be told that this person is the leader. After completing the task, group members will be asked to identify the member of the group that they thought carried out the role of leader. This will help us to identify individuals that emerged as leaders in the first condition and it will help us to verify that the assigned leader in fact took over this role in the second condition.

As mentioned above we will be using the SLP framework to identify linguistic features that can predict an individual leader amongst a group. In addition to analyzing the transcripts for linguistic features, we will analyze them at a higher level to identify the ways that leaders manage their presence and promote trust and cohesion through language.

Our work will contribute to the discussion of emergent leadership language, as well as how language is used to promote presence, trust, and cohesion in virtual leadership. This is relevant to the CHI community as it will suggest ways in which we can support the emergence of leaders in virtual groups and assigned leaders through computer-mediated communication tools. We will provide support for emerging leaders through automatic detection of this dynamic and by providing tools for leadership tasks that are made more challenging when mediated by technology. Effective leaders have been shown to increase productivity, making tools to support virtual leadership an attractive research area. There are many related areas for future work; we could run a study to investigate emergent leaders in asynchronous communication mediums such as e-mail, wikis, or forums. This work would have applications to online learning where students must collaborate, but may not have the benefit of face-to-face meetings.
In order to successfully prepare a short paper meeting the deadlines set by both CSCW/CHI and the course, we propose the following timeline:

• March 10: Submit IRB form
• March 22: Complete creation of all study related materials
• April 19: Complete data collection
• April 28: Complete data analysis
• May 5: Complete draft of paper
• May 12: Complete final paper
• May 12-August/September: Ask for additional feedback and revise as necessary


References:
1. Bell, B.S., Kozlowski, S.W.J. (2002). A typology of virtual teams: Implications for effective leadership. Group and Organization Management, 27(1):14-49.
2. Balakrishnan, A.D., Fussell, S.R., Kiesler, S. (2008). Do visualizations improve synchronous remote collaboration?. CHI ’08.
3. Cascio, W.F., Shurygailo, S. (2003). E-Leadership and virtual teams. Organizational Dynamics, 31(4):362-376.
4. Hancock, J.T., Beaver, D.D, Chung, C.K., Frazee, J., Pennebaker, J.W., Graesser, A., Cai. Z. (2010). Social language processing: A framework for analyzing the communication of terrorists and authoritarian regimes. Behavioral Sciences of Terrorism and Political Aggression, 2(2):108-132.
5. Tyran, K.L., Tyran, C.K., Shepherd, M. (2003). Exploring emerging leadership in virtual teams, in Gibson, C.B. and Cohen, S.G. (Eds.), Virtual teams that work: Creating conditions for virtual team effectiveness (pp. 183-195). San Francisco, CA: John Wiley & Sons, Inc.
6. Zaccaro, S.J., Bader P. (2003). E-Leadership and the challenges of leading e-teams: Minimizing the bad an maximizing the good. Organizational Dynamics, 31(4):377-387.
7. Zigurs, I. (2003). Leadership in virtual teams: Oxymoron or opportunity?. Organizational Dynamics, 31(4):339-351.

Analysis of High/Low Status Emails

Most people who sit down to write an email to their best friend use different greetings, words, and sentence structure than those writing an email to a potential employer. As this week's assignment confirmed, people alter their language use depending on the status of the person they are writing to. Our group compiled 20 emails addressed to high status members, and 20 emails addressed to low status members. These emails were then analyzed for parts of speech, and semantics. Listed below are some of our key findings, and our interpretation of this data.

According to Aoki & Woodruff (2008), "many communicative acts have implications for the status of a relationship." After analyzing our data, we feel that our findings support this statement.

NNB - It is likely that we used more titles in the high status category because email is more formal when writing to someone of higher authority. There were no examples of using titles in our low status sample. This could be because when writing to someone of less authority, the email tends to be more informal.

CCB - There were seven instances of the word "but" in the high status e-mails, but no instances in the low status e-mails. We are not sure why exactly this is, but we noticed that in the context of the high status emails, "but" was used as a form of justification. For example, when writing to a professor, students may write, "I know you said this, but..." When speaking to a member of a low status group, we may not feel the need to justify our thoughts and opinions.

PPY - The majority of the instances where "you" was used was in the phrase "thank you" or "thank you for your __". This is probably because people tend to use more polite language when writing to someone of higher status. When "you" was used in low status emails, it was more common to address the receiver as "you."

IW - The low status emails in our sample used the phrase "with" more than the high status emails. This may be because the instances of "with" referred to teamwork and collaboration. People of higher status may be more likely to supervise rather than work on the same level, so there would not be as much of a need for coordination to meet.

P1 - A lot of the high status emails are about education because they were sent to professors. Among those that weren't sent to professors, many of them discussed education as a conflict. For example, an email to a prospective employer included a class schedule. We were surprised that education was not more common in low status emails, as this is a common thread among the majority of our contacts.

A7+ - Our data shows that emails addressed to high status individuals used "likely" four times as often as low status emails. Based on our data, it seems students used this to ask a high status member if a specific task would be possible, or if they would be likely to complete said task in a particular time frame. Low status emails do not shows this thread, as the questions and requests are more direct.

T1.3 - Many emails addressed to high status individuals contain time periods because they are referencing when an assignment will be completed, or when they are free to work. Time periods are often associated with deadlines and assignments, which may be why time was mentioned five times more in the high status category. In terms of low status time references, time may be important when deciding when to have a meeting or eat lunch with friends.

N5- - Very few little quantities were used in this data set, but they really didn't follow a pattern. There were 4 instances in the low status group, and 0 instances in the high status group. The instances in the low status group used phrases such as "none" and "little". In these contexts, the words were used to further explain something; usually a problem.

A14 - Exclusivizers and particulizers were not used very often in this group of data, but they were more common in high status emails. We are not sure why this is true, but we hypothesize that students writing emails to high status individuals may be more likely to use these words to better describe or clarify their main point to avoid confusion.

Coding Scheme for Euphemisms

McGlone and Batchelor define euphemisms as a linguistic tool employed out of the reluctance “to utter more semantically transparent terms for certain unsettling topics”. In other words, a euphemism is an expression that refers to something in a more polite way instead of its literal name. Common usage of euphemism occurs with sex, pregnancy, profanity, bodily discharge, etc.

The following instructions must be abided in order for the coding of euphemisms to be successful.

• Details about the e-mails are to be ignored. This includes the sender’s address, the recipient’s address, the blind/carbon copy recipient’s address(es), the timestamp, and the subject.
• The salutation and the closing portions of the email must be disregarded.
• Except for e-mail details, salutations, and closings, each sentence in the body of the email is to be considered for euphemism.
• The sentence to be considered must be in English.
• If a sentence does not contain a euphemism, it is to be coded with the number 0.
• If a sentence does contain a euphemism, it is to be coded with the number 1.
• Every sentence must be coded as a euphemism with 0 or not a euphemism with 1.
• Each sentence must be coded once.

Once our team coded the e-mails, a reliability percentage of 91.4% was calculated.

A few examples of agreed euphemisms are listed below.

• so we're of age already.
• Work some team lead magic.
• haha he's smokin' here

Several examples of disagreed euphemisms are listed below.

• So success or epic fail?
• If we can leave them at home, I can anticipate several fewer moans and groans tomorrow at the beginning of class
• Do you know how else I would go about this?
• Thank you for your time.
• just give a text/voicemail/email so I have a heads up.
• My blank check awaits.
• AHHHHHHHHHHHHHHHHHHHHHHHHHH!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!111!!!!!!ELEVEN!!!
• Just a really minute detail

Language differences in the presence of shared workspaces

There has already been a great deal of research investigating the use of shared visual context while completing collaborative tasks such as solving a puzzle or a murder mystery. Most of this work focuses on video conferencing (so each participant can see every other participant), or sharing videos of each participant's real-world workspace (in the context of a puzzle, each participant can see the puzzle that every other participant is working on). There is a smaller body of research that looks into shared digital workspaces for collaboration tasks, and a lot of this work focuses on heavily constrained work spaces that have been developed or adapted for a particular task. We are interested in how the presence of a less-constrained shared workspace such as a digital whiteboard affects the way people interact with others. We are also interested in how people adapt their language and behavior if the medium they are using is not conducive to the task they need to solve. Specifically we are interested in how different group and goal-oriented tasks are more appropriate to be completed via text, a shared visual workspace, or a combination of the two and how groups compensate for a poor match between task and communication medium.

In order to study our question, we plan on using a 3 by 3 factorial design. There will be three forms of media richness: chat only, digital whiteboard only, and digital whiteboard with chat. There will also be three tasks that are yet to be determined. In choosing tasks we plan on choosing one task that will be easiest under the whiteboard condition, one that will be easiest under the chat condition, and a third that will be easiest under the mixed condition. Under these conditions, we will observe the similarities and differences in the language used and exchanged between the participants of our study, their use of the whiteboard, and how efficiently they completed the task. Our team hypothesizes that the language used will vary within the different spaces and that certain tasks are better suited to be completed under certain conditions.

As our team continues to solidify our research question and procedure, there are still matters that are uncertain and yet to be clarified. While we believe that dictating street directions to a person would be most feasible via an on-line whiteboard and telling a story would be most appropriate via chat, we need further guidance in selecting our tasks. Also, while our team members have shown interest in this research space, we are still trying to answer the question "why?" We want to ensure that we can make practical recommendations based on the findings of our study. We question if finding differences in language usage under the three noted conditions will suggest any kinds of modifications in present day conventions of groupware and collaborative tasks.

Email Etiquette

Research Question
Conversations can be divided into three segments: an entry where two people agree and coordinate the commencement of a conversation, the body where the participants discuss one or more topics, and the closing where all conversation partners agree to close the conversation. In asynchronous conversation mediums such as email, conversation boundaries can become blurred when one full conversation can be composed of a number of discrete messages that themselves can be made up of an entry, body, and closing. We are interested in how the composition of the individual messages changes over time in a prolonged email exchange. More specifically, how do people alter their greetings and signatures within prolonged email exchanges?

We hypothesize that greetings and signatures will be altered to mirror conversation partners within prolonged email exchanges. In order to test our hypothesis we have designed an experiment where we will track a thread of emails between one subject and one confederate in two conditions, and two subjects in the third condition. Regardless of condition, each pair will be given a task, which they must complete through email.

Condition One
The subject, an undergraduate, will be given a task and a partner (the confederate who they are told is also an undergraduate) to complete the task with. The subject must initiate the email conversation, but after that, anyone can write at any time. In this condition the confederate will stop using greetings and signatures on the third email that he/she sends.

Condition Two
The setup and participants are the same as in condition one, however in this condition the confederate will use greetings and signatures throughout the exchange.

Condition Three
In this condition both subjects will be undergraduates. They will be given the same task and told to complete it via email.

Further, we hypothesize that the first message from each partner will include a formal greeting and introduction, as the partners do not know each other. In condition one we hypothesize that once the confederate changes his or her entry method, the subject will follow in order to take part of the joint project. In the second condition, where the confederate continues to keep the entries and exits formal, we hypothesize that the subject will do the same in order to be polite.

After collecting the dialogues, we will throw out any dialogues with less than six discrete messages and those that took longer than 2 days from start to finish. After reviewing the transcripts we will make a determination as to what constitutes a greeting and a signature within a message. We will then code all of the resulting greetings and signatures on measures such as length and formality. Finally, we will use our coded data to observe patterns in the greetings and signatures of conversations between conditions.

The Now&Later Pledge

With the members of Team Now & Later as my witnesses, I pledge to do the following things in order to ensure success throughout the duration of this project:

1. I promise to do what I say I will do.


2. If circumstances arise that prevent me from doing my part, I will notify my team members in an appropriate manner.


3. I will keep my team members informed of my progress and update them on my status.


4. Above all, I will keep team morale high!