To find data regarding the linguistic differences between adolescent males and females in online communities, I used primarily passive observation. To do this I made an account on Habbo Hotel. This entailed things like designing an avatar and username, both of which I attempted to make nonobtrusive to ensure online users would not pay my present-but-silent avatar much mind in the online chat rooms.

To make my observations on Habbo Hotel, I had to decide what kinds of chatrooms I wanted to go into since Habbo offers a wide variety. These rooms can range from official introductory rooms for new members, to eighteen and older chat rooms, to swap meet for virtual possessions, to money making ventures for those seeking to earn virtual currency. For the purpose of eliminating variables in my data, I decide to exclusively use official chatrooms sponsored and created by Habbo Hotel. The rooms I visited most often to observe conversations were the park, the movie theater, the club, and the coffee house. I alternated between rooms and used the navigator function to see how many people were in a given room so that I could choose one that was busy enough for me to easily find data, but vacant enough not to impede my observations. I  chose not to take any data from the “Welcome Lounge” room despite the fact that it is an official Habbo room because I thought that people new to the game would likely display different linguistic traits than those who had been in it for a while and understood the status quo. I did this for the sake of the integrity of my data. For example, new people may tend to talk less than they would normally because they are unfamiliar with how the chat function works or perhaps they would be more profane than normal because they are unaware of the game's censorship capabilities.

Each time I went to make an observation I made a deliberate choice as to which room I went into. I almost always chose from the park, theater, or cafe because not only were these rooms Habbo official but they generally ranged from 15-45 people present which is ideal for the purpose of my observations. I made it a point to alternate between the rooms I went into the minimize the risk of observed the same individuals over and over again. In addition alternating rooms, I made it a point to record conversations that took place during different times of the day so I was not, for example, getting primarily east coast users or only British users or something to that extent.

When it came time to actually making observations, the process was simple. I entered the chat room and stood somewhere nonobstructive. All of the conversations are visible to everyone in the room, the bubbles appear above where the avatar who said them, however, this view scrolls and disappears and was not very conducive to counting out words and recording data, so I used a more simplified view available and took screenshots to record the data. I generally recorded ten to fifteen minutes worth of information.

Once I had the data I needed I went back and reviewed my screen shots. To record the data in a more functional form I used excel to recorded the username and displayed gender of the user. In subsequent columns, I recorded the number of lines of text submitted, the average number of words per line of text, and the total number of words. In addition, to that I counted the number of emojis, laughter representatives (lol, haha, hehe, lmao, etc), substituted profanity (fck, a s s, bobba, etc), and completely capitalized words. I then looked at how males vs. females compared proportionally based on the male to female ratio in the conversation and how often each gender did the specific things I recorded. In totality, I collected data from thirty conversations and 174 total users. Of those users 85 were male and 89 were female.

There were several challenges I faced in deciding on the methodology of this project. One problem that may be consequential for my data is the fact that conversations never really end on Habbo Hotel. Users enter and exit rooms frequently and may or may not choose to participate in the chat room’s dialog while there. This means that in the ten to fifteen-minute observation window I observe from, I could be catching just the last three lines of text from someone who has been talking for hours before they log off- or contrarily I could catch the only five minutes that a user decided to speak in the chat room. This is a problem symptomatic of randomly selected observations, however, it is my hope that having thirty different conversations to derive data from will even these issues out.

Another difficulty worth noting is the censorship on profanity the site uses. If a user types out a curse word and hits send, the curse word gets replaced with the word “bobba.” I decided to treat bobba like I would treat “a s s” or “fck” because of the profane intent behind the usage despite the fact that it is censored and I have no way of knowing what specific curse word was used.

The final challenge I faced with my methodology was that at times it was difficult to determine the gender of a user because at times they are presented as being highly androgynous. Although most of the time it was easy to determine the gender represented by the avatar thanks to characteristics like breasts, or short hair, or masculine clothing; avatars are highly customizable, so users wearing things like paper bags over their heads or Halloween costumes made it a difficult to determine whether a male or female was being represented.