What Does That Text Mean?What Does That Text Mean?

Text tone analysis could lead to improved understanding and better interactions.

The contact center has become the interaction center. It still supports traditional voice conversations, even with the addition of support for text, chat, and conferencing channels. If trained and experienced, agents should be able to grasp the emotional states of callers, but when text, chat, and IM are added into the mix, it becomes harder to discover the emotional state of the caller.

Does the agent wonder what the caller is really conveying via text? Can emotion be detected through emails, texts, chat, and IM? Tone measurements could be added as a mechanism for evaluating agent/customer interactions. It can even be used to measure the tone of text collaborations and conferencing.

A text tone analyzer computes language tones by performing linguistic analysis. It is done by correlating the tones and linguistics features that are found in the text. There are three forms of tones that can be analyzed and measured.

Emotional tone is derived from performing emotional analysis on a text statement and/or interchange. This can be scored. Measurements of opening greetings, punctuation, emoticons, curse words, and other factors can be fed into machine learning algorithms that can classify the text into emotion categories.

There are five personality characteristics. The article, "The Personality Insights models" goes into much greater depth discussing the following five traits:

Language tone can be calculated and measured from linguistic analysis based on learned features.

IBM has launched a Generally Available (GA) version of the Watson Tone Analyzer. You can check out a demo to get an overview of the service, or try it out for yourself and build your own creative Tone Analyzer applications.

The GA version includes service improvements that enable tone detection in a broad variety of communications. It incorporates additional features such as punctuation and language parameters, providing more robust tone insight. The emotion detection models are trained on large data sets that include features such as emoticons and slang words. You access and use the service through APIs.

The IBM Tone Analyzer service is based on psycholinguistics that explores the relationship between linguistic behaviors and psychological theories.

People produce various language tones in their everyday communications. The emotional tone can be joyful or sad, open or restrictive/conservative, analytical or informal. The tones can be read correctly or misinterpreted. If a person speaks in a monotone, it is difficult to decipher their emotional state, and that can influence the entire interaction adversely. This impacts a person's perception of the identity of the other person. It can improve or disable the effectiveness of their text interactions.

People perceive negative emotions produced by senders with stronger intensity than they do positive emotions. Emotions can be mistaken as neutral when in fact they are positive. The personality characteristics that people express also influence others' perceptions of their occupational proficiency. Are they competent or not? People present different online identities in social media that impact the impression that others have of them. The measurement of the interaction can be scored as shown below.

There are many possible uses for tone analysis, but here are some of the top ones that come to mind: