Johannes Gutenberg-Universität

1.1 Introduction

As a frequent user of the internet I have discovered various phenomena, which caused me to contemplate their purpose and effect on the people behind the keyboard. So far I have refrained from researching those phenomena in existing works about the language on the internet, mainly to allow myself to form my own ideas first. But I am aware that any research beyond the point of this pilot study will need a thorough research of the linguistic works on the field. I am also aware that I might have invented the wheel anew by abstaining myself from the available literature, only that it is just a very small wheel, if one at all. But my main purpose of conducting such a field study was to learn some of the techniques that may be useful. Also it may give answers to some of my questions.

The internet provides a number of ways to communicate with other people in the world, such as e-mail, news-boards, web-pages, video-conferencing and chatting. The selection is exemplary but not exhaustive of all the possibilities on the internet. While all of the named forms are a means to communicate your ideas to the world and receive answers from other users of the internet, only video-conferencing and chatting are conducted in a synchronous way; they happen while two or more users are active on the internet, sending and receiving messages synchronously or in terms with no, or only little temporal breaks in between. Of the two video-conferencing resembles an ordinary conversation between people in so far that they can see and hear each other, and also see the reactions of the others and hear their voices, while these elements are completely absent from a text-based cha”. In order to establish a video-conference the participants also need a certain amount of technological equipment, knowledge of the equipment, and monetary resources to provide for a ‘good’ link. A fact which limits the use of video-conferencing to few occasions in comparison to the millions who have access to chat-providers through simple programs which usually are either free of charge and can be easily installed from the web, or which are purchased together with the hardware (computers) the users buy. Therefore it is much more interesting to investigate a relatively ‘new’ form of conversation that resembles neither the spoken language nor the written language completely, and which is used for communication by a multitude of people all over the world and every day.

1.2 Internet-chat

There are mainly two types of internet-chat; one of them was supported by a Unix-based program called ‘talk’. Today it is not used much any more and institutions like the Johannes Gutenberg-University of Mainz for example have blocked it completely from the internet because of security reasons. Here you could see what the other person was typing, while (s)he was typing. And you also saw when (s)he made mistakes and corrected them. The other kind of chatting is based on turn-taking and it’s the more known and prominent form nowadays. You can find it supported by programs such as kIRC (Linux based), MSN (Microsoft/ Windows based), or ICQ for example. The programs link the user to common providers which are located on the internet and which are run by various institutions, such as universities or companies. In this form the chatter writes a message and sends it to the recipient(s) when (s)he considers the message as complete. This allows everyone to send messages at the same time, while the displayed messages appear in turns and sorted by the absolute time when they were received by the chat-provider. Due to modern technology it doesn’t take much more than a second for a message to be relayed to the other participants. Of course there are exceptions when the internet is slowed down, but they are infrequent and they don’t play a role in the research I conducted.

The purpose of chatting ranges from private flirts over social gatherings to serious discussions about virtually any topic of interest to the participants. While it is possible for many people to participate in the same conversation, for example in a chat-room, I want to limit my observations to conversations between two people. I am aware that a conversation that is limited to two people most probably has a different potential from a conversation between more than two people, but the larger setting also causes more social interferences, and I would have to keep group dynamics in mind as well as my own goals.

1.2 Elements of the chat language

One of the most prominent phenomena, while chatting with someone on the internet, was the fact that sometimes a normal sentence did lead to major misunderstandings and in some cases even to an agitated discussion about the statement, and it’s meaning. Most of the sentences were not different from anything else that had been written, yet they caused very strong reactions. After a while I started considering the circumstances a bit more deeply.

One example for such an occurrence was the sentence:

(1) 19:10:25 CurrentUser: “I’ve got to finish a translation”.

This was sent in response to a question on what I was doing. In a normal letter (or e-mail) the sentence would have to be considered in light of the circumstances of the letter, as well as the complete content, and the fact that the letter had been written at all. In a ‘normal’ conversation the sentence would have been spoken in a certain tone, and the recipient would have been able to see my face.

I paid closer attention to the way I had formulated the sentence as well as to my own condition. I was tired and somewhat preoccupied by the translation I was working on, and I found that I had not attached any of the usual elements to the sentence; elements which are used in the chat-language to indicate the mood of the chatter as well as activities of the participants.

There are mainly two categories of visual elements in the chat language, which are clearly distinguished from the rest of the written language by their nature. The first are the smilies, which initially were produced by placing a colon and a closing bracket next to each other like this:

It’s supposed to be the graphic resemblance of a smiling face. I can’t remember seeing any other kind of smilies for a while after I had seen the first, but at some points more of them were introduced with varying meaning. While I am not sure if there is any other means to distinguish the meaning other than by either asking or trying to ‘see’ what they look like, I am fairly certain that not every existing smiley is understood by each user of the chat-language. Nevertheless for many of them a fixed meaning can be determined, which probably will be understood in the same way by the majority of the chatters. I will introduce two more, which were used in my pilot study:

;) - a twinkling face

:( - a sad or disappointed face (this is still called a smiley, even if it doesn’t

smile)

The smiley has been anchored so deeply in the chat-language that programmers of chat-programs as well as text-programs such as Microsoft Word have integrated it into their repertoire, and using :) will usually produce this: J

There are other forms of smilies, such as ^_^ for example, but they belong into the same category of graphical display as the classical smilies. I also found that a number of people from Australia that I have spoken to produced their smilies facing the other way, like this: (: which is something that I found very interesting, and which I might research at another time.

Sometimes other ‘items’ are also ‘drawn’ with ASCII graphics, for example such a rose: @-`-,-}-- But they are less frequently used, and their meaning seems to differ from the smilies in so far that they are meant as a ‘gift’ to the other participant and not as an indicator for the mood of the chatter.

The second category of elements are the actions or emotes. They resemble the actions of the chatter during the conversation. There are different ways to set them apart from the rest of the text graphically, such as putting them between asterisks; *action* or between special brackets; <action> (this kind of brackets is used to mark tags in html – the hypertext markup language, which is used to describe web-pages). There may be other ways to mark actions, but those two are the main forms which I have found in usage so far. I think the way of marking the actions is probably derived from computer languages, but I have not confirmed this, yet. A typical sentence or message complete with action and smiley could look like this:

(2) 20:30:07 Name: I’ve found a new pic. J *hops around*

This is an example I have taken from a conversation on ICQ. The numerical part indicates the time when it was received, followed by the name of the participant, which is usually a pseudonym (I have changed it here, to guarantee privacy to my testers), then the message, a smiley and an action, which is supposed to give us a mental image of the other user hopping around. I found that actions often are exaggerated like in my example. They tell us how the other feels. This user felt like hopping around, even if (s)he did not actually do it. I’m not sure if that is the reason why actions are also called emotes, since they often describe the emotional feelings of people, or if the name emote derives from the use of a command called ‘emote’ which was used in specialized chat-environments called MUDs or MOOs. However that is not essential to my study either.

The actions have been integrated into some of the chat-programs as well in such a way that a user can initiate a message with a special command which will change the layout of the message. A possible result of the earlier example would be this:

(3) 20:31:15 Name hops around.

The time-stamp and the colon are usually generated automatically by the chat-program. In this example the colon has been removed by the chat-program, due to a special command of the user, so that a full sentence can be formed, starting with the name of the participant who sent the message.

So when I sent the message (1) I was tired and busy, and I forgot to insert a smiley or an action to explain my mood, or to disable any of the possible meanings of my sentence, which would have been obvious in the setting of a letter, or in an actual conversation.

My chat-partner explained to me that I didn’t have to talk to him, if I didn’t want to, and that he was annoyed. It turned out that he had read a second meaning into my sentence, one that was not necessarily intended, namely to leave me alone since I had more important things on my mind.

Inspired by this experience and many others which turned out in a similar way I decided to inquire more deeply how users of the chat-language perceive these elements. And it is with those facts in mind that I started to design my tests.

2.1 Designing the test

My first idea was to ‘face’ people one-to-one on the internet and provoke reactions by either using none of the elements, or asking them to refrain from using them. While I was looking for topics and possible phrases, which could be used to provoke a reaction from my testers, I became aware of several facts:

First of all I needed to find testers who would spend at least fifteen minutes of their precious time with me in a chat-situation which might not be comfortable for them, especially not if I wanted to provoke a misunderstanding between them and me. Of course I could have asked my friends to help me out, but considering that I am planning to do a full study later on, such a test would not work well with people who don’t know me at all. And then my research would have to be limited to few testers.

Secondly I needed to arrange for them and me to be online at the same time. This is very time-consuming, and it needs a lot of understanding from the other side as well.

And finally both my testers and I would be subjected to a situation which might not create the usual reactions, since I had planned to inform them of the testing situation, and that I was creating a log-file from our conversation.

Taken together these facts seemed to stand in my way of producing the natural reactions I was looking for, which led to the second approach, a different kind of test; I logged a conversation between a friend and me. My friend knew about my research and I asked him not to use any actions or smilies in the conversation. Despite that he slipped once and produced an action, and it took both of us some conscious effort to avoid the chat-elements altogether. In this way we produced a real conversation using ICQ.

I chose the topic of the conversation by the field of interest which I expected my testers to have. In this case it was a discussion about the movie ‘Shrek 2’, which my friend and I had watched together.

In the next step I created two tests from the logged chat; I replaced the names with ‘Person B’ and ‘Person C’, I didn’t want to use ‘Person A’ because I thought that it might lead people to read it as ‘Persona’, and the –a Suffix is frequently used for female names in most languages who have Latin influences. This may have proven to be a bad choice, since real names are much easier to remember and assign attribute to. I suspect that some of the testers might have been confused by this, but I can’t be certain. It is a point which I will have to keep in mind when I run my next test.

In the next step I inserted smilies and actions into both tests. In ‘test A’ I had only ‘Person B’ use the smilies and actions, while in ‘test B’ they were used only by ‘Person C’. I didn’t alter anything else, which may have proven a mistake, since it did influence the results. Both my friend and I usually use full sentences and proper punctuation, but since I have seen many chatters start their messages with lower case only, as well as creating only half sentences most of the time, I had decided to imitate that style myself, but I had forgotten to tell my friend to do the same. This created an effect which influenced the reactions, which was also stated by some of the testers, and which I will explain together with the results.

Following the tests I put several questions, the first set was about the personal status of the testers; they consisted of age group, gender, location on the globe, first language learned, education, and current social circumstances. At the question for gender I asked them to put their perceived social gender, not necessarily their natural sex. The next set of questions was to categorize both ‘Person B’ and ‘Person C’; they asked for the perceived gender, age, likeability, and to categorize them into a given number of attributes as well as to give reasons for their choices, if there were any.

The choice of the actions and smilies proved rather difficult, since they had to fit in with the content of the text, and I ended up using both ‘positive’ and ‘negative’ elements, but I tried to use the ‘positive’ ones more often.

2.2 Running the test

Picking my testers proved less difficult since I am part of a very productive internet community where I found several willing and helpful people. Half of them I asked personally, my friend helped me with those as well, the others I had asked over a public mail to approach me if they were interested in contributing to my study.

In turns I sent either ‘test A’ or ‘test B’ to the volunteers, asking them not to compare their test with the others until they had finished it.

The test was a in form of simple text file which can be viewed on any computer system, nevertheless I received one e-mail asking if the formatting was supposed to be that way. It seems that this person had used a viewer that did not format the text properly.

Another question I received was: “Where is Person A.” And all it all it may have been a bit hard to keep in mind which person had said what in the text. Again this could have been avoided by choosing real names that are not gender-specific in any country. Also the text could have been shortened a little.

The whole test was rather lengthy because of the length of the chat-part. This was good in such a way that I wanted the first reaction from my testers, so they would not go back to the chat and compare it once more, but at the same time it was bad, since my choice of synonyms for the two chat-participants to be evaluated by the testers was contra productive.

2.3 Evaluation of the results

I sent ‘test A’ to twelve people in total and ‘test B’ to 12 people. So far I have received 11 replies on ‘test A’ and 9 replies on ‘test B’. I might still receive replies from the other testers, but sadly I won’t be able to include the results in this paper.

I am aware that the number of testers is by no means high enough to confirm a theory, but it may point towards tendencies which could be confirmed in a larger worldwide research with a much higher number of participants and an optimized test.

Nevertheless I would like to point out a number of tendencies which can be deducted from my results.

After gathering all results in one table I sorted it by test and found that in ‘test A’, where Person B had all the smilies and actions, while Person C had none, 6 people thought that Person B was male, and five people thought that Person B was female. Also 5 people thought that Person C, who had none of the elements, was female, and 5 thought Person C was male, while one could not decide. (See table 1.1.)

In ‘test B’ there were 7 people who though Person B was male, and two decided for female, while 7 people thought Person C was female, one thought Person C was male, and one did not fill out the question. (See table 1.2.)

Table 1.1 Table 1.2

Test A	Person B	Person C
Male	6	5
Female	5	5
Other	-	1

Test B	Person B	Person C
male	7	1
female	2	7
other	-	(-)

In reality Person B was female, and Person C was male.

This leads to the question if there is a tendency for people to perceive the combination of smilies and actions together with well formed sentences which begin with capital letters and end with proper punctuation, as a female characteristic.

My next step was to check the attributes against the results, and I found that in test B two people found Person C grumpy. One of them was the tester (TP24) who thought Person C was grumpy, and (s)he was also the same who thought Person C was male and Person B was female. This could indicate a possible confusion about the identity, but I prefer to believe that my testers knew what they were doing.

I categorized the attributes into groups of awake, tired, interested, disinterested, negative (which contained bored, grumpy, unfriendly, gloomy, brooding, and sad), and positive. The other categories I considered as rather neutral. (See table 2.1.)

Table 2.1

	A - Person B	A – Person C	B – Person B	B – Person C
awake	4	3	1	2
tired	-	2	-	1
interested	4	5	7	5
disinterested	-	-	-	-
negative	2	3	3	4
positive	11	8	2	6

This led me to two careful assumptions: 1. My testers were aware of the test situation, and the majority preferred giving positive attributes to Person B and Person C.

2. There might by a tendency that smilies and actions are perceived as signs of a positive and active state of mind, while most testers were rather cautious with assuming negative attributes from the lack of such elements alone.

I also received comments that the existence of the actions and smilies pointed towards a younger age. (See table 3.1.)

	A – Age B	A – Age C	B Age - B	B – Age C
young	5	5	-	3
middle	6	5	9	6
old	-	-	-	-

Table 3.1

While 5 people considered Person B as young in test A, none of the testers of test B thought so. This could indicate that people do relate a high emotional activity with young people. But at the same time the results of Person C indicate that the presence or absence of activities and smilies alone is not interpreted in such a way. Therefore it is necessary to eliminate all other differences in the used language for such a test to be useful.

3.1 Conclusion

Including age and gender of the testers, as well as their background and comments into the study may lead to further interesting results, and it may have been a major factor in my test as well, but the actual numbers that I have gathered now also has showed me that I need a much larger group of testers for the kind of confirmation I am looking for. I had twelve female testers and eight male testers, and the tests were divided equally among them. It may be interesting to see if male testers are less likely to react to the absence of smilies and actions than female testers.

In addition to increasing the number of testers my test has to be refined and optimized as far as possible in order to eliminate any possible interferences. A neutral and shorter text has to be found which allows little or no conclusion about the chatters.

For the applicability of such a test to a huge number of users it might be favourable to create a web-page on which users can fill out a test-form which will automatically be sent to me afterwards. The location of the web-page could be distributed through other universities.

Considering the gathered results I have to say that even if there may be tendencies visible, they are by far not strong enough for me to draw any certain conclusions. But they have inspired a number of ideas of how to improve my technique. And they have brought up new directions into which further investigations could be made.

Sources:

I have not used any written sources for my research other than the chat-examples which have been taken from my own logged chats with other people, who consented to my use of them. My knowledge about the internet and internet-security are based on a Linux-Course which I have participated in, and on my studies in BWL where my focus is on IT, both at the Johannes Gutenberg-University, as well as on my own experiences on the internet.

Comparative Linguistics

Conducted by: Prof. Dr. A. Sarhimaa

WS 2003/2004