Full version: The Limitations of Text as Contributors toward Miscommunication and Reciprocal Incendiary Comments

Here's the full paper for anyone who was interested. Sorry for the odd formatting.
Or if you were directly linked here and just want the bits I pulled out, here's that.

The arrival of the Internet and online communication offer great potential for society. On a global scale it offers the possibility of communication across national boundaries, connecting people who would otherwise be separate. The anonymity of it means that persons who might otherwise refuse contact due to ethnic divides or otherwise, can meet and discuss ideas in a medium free of bias and borders. For already existing social groups, whether they are friends or coworkers, electronic communication offers the ability to communicate from almost anywhere. This allows networks which might otherwise be severed or strained by distance to retain connections. While phones already offered this possibility, once an internet service has been paid for, the potential volume of email communication is nearly limitless.


However this enlightened online utopia has failed to materialize. Instead a new form of conflict has emerged: flaming. Exact definitions are hard to pin down, but a general consensus is that “flaming consists of aggressive or hostile communication occurring via computer-mediated channels.” (O'Sullivan and Flanagin, 2003). A flame war emerges when attacks are reciprocated and can even involve additional members. Flames can take many forms with no limits to their absurdity or lack of basis in fact. Part of this is due to the inability to verify the accuracy, or more often, inaccuracy, of flames. The same anonymity which allows for anyone to freely express an opinion also allows that individual to express negative or hateful opinions.

I will focus the effects of limited communication and how it may contribute to flaming. This is in contrast to theories which suggest that flaming or trolling (another incendiary behavior which often overlaps) are caused by personal factors. My goal is not to disprove these theories, but instead to look at another factor. While personal factors are important, the environment is more easily changed, so I believe that a focus on the limits of text offers more practical value for prevention of flaming.

In face to face communication, misunderstanding may be readily apparent. This allows it to be corrected. The correction begins with the speaker noticing that the listener misunderstood; either by failing to draw the same information as the speaker intended or by receiving nothing at all. Noticing can be a simple matter of seeing a confused expression, a non-verbal and automatic form of communication. Verbally, the speaker may notice a disconnect between expected feedback and actual feedback. The speaker can then attempt to correct the misunderstanding. The correction may or may no be understood by the listener, but through an exchange of words, non-verbal cues, and feedback on each, the two individuals, or a larger group, can eventually sort through the misunderstanding (Bazzanella and Damiano, 1999). This process is more likely to fail in online, text-based communication. It is this failure to notice and correct misunderstanding, along with increased chance of misunderstanding, which set the stage for flaming.

To understand the sources of misunderstanding, it is necessary to know the limitations of text and how it differs from face to face communication. There are no non-verbal cues in text; this means no facial expressions and no posturing or gestures. In addition, there is no tone of voice. To be more accurate, there is no defined tone of voice, instead the reader uses whichever tone he chooses, regardless of the tone heard by the writer. Timing is different as well. The writer cannot use pauses or changes in speed to modify his message. Messages are more spread out as well because rather than formulating a response while listening, the reader must wait for the entire message to be sent before he can respond to it. The difference in timing also means that non-verbal cues such as long, awkward pauses are much less noticeable. All of these limitations restrict the way in which information can be transmitted, increasing the likelihood of a failure to communicate.

Despite the limitations of text, both writers and readers frequently overestimate the completeness of their communications. Writers overestimate their ability to transmit sarcasm (Hancock, 2004), causing the reader to receive information which is the opposite of the intention. Because of this overconfidence, writers fail to compensate for the limitations of text and readers assume they understand (Kruger, Epley, Parker, Ng, 2005), together eliminating the feedback cycle which detects and corrects miscommunication. Writers may use too familiar of a tone, communicating as if there was greater mutual understanding than actually exists. This is seen often in emails which tend to be short and assume familiarity with the reader (Baron, 1998). In addition, while profanity is held to be an indicator of flaming, it does not always carry the intent. However despite the lack of intent, profanity may be interpreted in a hostile manner if unaccompanied by mitigating context or mutual understanding (Turnage, 2007).

Sarcasm can easily create the impression of disagreement. Failing to detect sarcasm will lead a reader towards a literal interpretation of text, causing a contradiction of the intent. If they lack previous experience the reader may not recognize that the statement is unusual for the writer and will instead base replies off an incorrect start. This can be especially conducive to emotionally charged flame wars since sarcastic statements may not only contradict belief but also be completely absurd, as I have personally experienced. On the other side, sensing sarcasm where there is none can cause two problems. The first is contradiction of intent as seen with failure of detection. Sarcasm may also be interpreted as rudeness, so a false positive creates a mistaken perception of rudeness on the part of the writer (Byron, 2008).

Humor and its failure is similar to sarcasm and its failure. Failure of detection will lead to misunderstanding between the reader and writer. Both humor and sarcasm detection failures are promoted by the lack of non-verbal cues in text. The smile which would indicate a joke is absent in addition to the pacing required for proper telling of most jokes. Because the writer does not provide a voice, jokes can end up with a dull delivery, leading to a gap in perceived humor Kruger, Epley, Parker, Ng, 2005).

Even if the reader does recognize and enjoy the joke, when responding he will become the sender and will be subject to the emotional limitations of text. There are several trends of failure when transmitting emotion through text. Due to the lack of voice or facial expressions, messages which are intended to carry emotion can be perceived as having neutral emotion (Byron, 2008). This causes praise to seem muted or non-existent. In extreme cases, praise which does not carry emotion may be perceived as sarcastic. These tie in with negativity effects. In face to face communication negative information can be tempered with consolatory gestures such as a hand on the shoulders or expressions of empathy such as matching of facial expression (Derks, Fischer, and Bos, 2008). Lacking these moderating gestures, negative information from the writer to the reader can be perceived as personal attacks.

At this point the communication between writer and reader is not entirely lost. Having received feedback, the writer has the opportunity to recognize his failure to clearly transmit moderating emotional information. However overconfidence suggests that he is more likely to think his message was understood and that he is the one being attacked, escalating the conflict. Alternatively, he may exhibit a positive affect bias, failing to recognize the emotional distress. This tendency has been found in comparisons of text and video expressions at cancer support groups (Liess; Simon, Yutsis, Owen, Piemme, Golant, and Giese-Davis, 2008). The reader may continue with his side of the discussion, oblivious to the miscommunication and therefore failing to take corrective measures to stem the tide of negative emotion. In a face to face encounter the imitation of expression combines with self-moderation based on social queues to limit socially unacceptable anger. Without these moderators, anger in text-based communication can spiral out of control (Derks, Daantje; Fischer, Agneta H.; Bos, Arjan E. R, 2008).

Anonymity causes many problems for communication. The lack of previous experience makes it difficult to predict responses from readers. Similarly a lack of background information makes it difficult to detect social norms (O'Sullivan and Flanagin, 2003), leading to unintentional rudeness. However, reducing anonymity may only trigger use of misinformation. The reduced information in text leaves more gaps to be filled. Comparing phone and email communication it was found that small amounts of cueing information, such as race, heavily influence perceptions of email messages, even with the exact same words. Small amounts of identity information allow for stereotypes to be activated and fill the gaps (Epley, Kruger, 2005). The irony created is that the potentially race-blind online world of text can instead become focused on nothing besides race and potentially negative stereotypes.

Other forms of stereotyping can come from the writing. Poor spelling can carry many inferences: the writer doesn’t care, the writer is lazy, the writer is stupid. Grammar carries similar potential labels. It is another irony in the online world that a poor speller is criticized for poor communication while the reader activates a set of preconceptions which may be entirely inaccurate and certainly not based on any known qualities of the writer. This is a clear example of how understanding is based not only on the writer’s intention, but ultimately on the reader’s interpretation (Byron, 2008). In the middle of a flame war though, no one is safe, and so the readers who give feedback based on spelling or grammar may be labeled Grammar Nazis. This does not elicit kind responses from the labeled.

Together all these limitations and miscommunications contribute to the formation of flame wars. The non-verbal cues which normally add emotion and subtly to speech are absent and writers often fail to compensate (Kruger, Epley, Parker, and Ng, 2005). Missed or unintended sarcasm can cause perceived disagreement, and while contradiction is insufficient for a true argument, it forms the foundation of a flame war. Failing to convey emotion leads to mistaken personal attacks and if the writer does not notice the hostility he may continue with the content or writing style which triggered the negative reaction. Stereotypes distort the perspectives of readers, causing them to twist the words of the writer to confirm their beliefs, regardless of content. The overall effect of the exchange is a reduction of whatever respect may have previously existed and “promote widespread depersonalization” (O'Sullivan and Flanagin, 2003).

Despite the limits of text, it is possible to compensate partially for its limits. Adding clarifying words, using fewer ambiguous words, employing emoticons, and monitoring tone from an outsider’s perspective will improve the writer’s ability to deliver context along with content (Turnage, 2007).

Emoticons attempt to mimic human faces as a way to convey information. They are limited in their range of expression and beyond simple smiles and sad faces become harder to interpret, reintroducing the difficulty of communicating emotion. In addition, emoticons are slow and deliberate while facial expressions appear almost instantly and without effort, making them weaker as accurate indicators. However, emoticons do still approximate the pattern of facial imitation as an expression of empathy (Derks, Daantje; Fischer, Agneta H.; Bos, Arjan E. R, 2008) and so can be useful for showing and moderating (through the feedback against excessive negativity) emotions. The adoption of emoticons is limited, within businesses they may be considered unprofessional despite their usefulness and are still limited in their purposeful use (Baron, 1998) (Byron, 2008).

Self-monitoring may be the most effective means of improving communication. Writers and readers can both improve their accuracy by rereading text in order to check the tone (Turnage, 2007). Reading out loud in a neutral tone has been found to reduce overconfidence in text-based communication. This prevents the writer from using his own mental voice to add his interpretation to the words, allowing him to better see how his words will be read (Kruger, Epley, Parker, and Ng, 2005).

Flaming is wasteful in many ways. While responses may be hastily written, they nevertheless take time and watching the progress of a flame war takes even more time. Attempting to moderate such wars can be time-consuming, if they are even noticed before they’ve wasted time. Outside the context of a flame war, electronic miscommunication can be damaging to relationships and productivity within business. Preventing miscommunication by altering the way in which text is sent and read would benefit everyone and increase the effective use of electronic communication in both email and online discussion.

Emoticons can increase meaning, but are limited by their expressiveness and ability to be interpreted. They may appear unprofessional as well. As an alternative way of adding tone and emotion to text communication, I propose a richer style of writing to be tested. Rather than leaving emotion to luck, the writer instead actively writes to convey emotion, using phrases such as “I am happy with ___.” Additionally, the writer explicitly states the intended tone and acknowledges the limitations of the text: “I do not intend for this to be read in a sarcastic manner, instead it should be read literally, as praise” or “my intended tone is positive, though it may fail to be conveyed by the blandness of text.” Repetition with modification would be employed as well, giving the same point in different ways in order to proactively imitate the rephrasing which is used to correct misunderstandings. Used together these changes will make text-based communication significantly longer. This is actually a good thing as it will help to change emails or discussion posts into a more deliberate form in which there is more time to consider one’s words and potential reactions from readers. While readers might be impatient, they would find it worth the tradeoff when they, as they perceive them, receive more positive and less ambiguous information from the writer.