MANY FACES OF A CHATBOT: THE USE OF POSITIVE AND NEGATIVE POLITENESS TRATEGIES IN ARGUMENTATIVE COMMUNICATION WITH A CHATBOT

: With the ever-growing development of artificial intelligence,

communication and argumentation processes are no longer limited to face-toface and computer-mediated communication with other humans. They now

involve communication with machines. This fact has given rise to chatbots,

computer programs trained to produce human-like communication through the

use of highly specified algorithms. The aim of this study was to determine

whether a chatbot has been trained in accordance with positive and negative face

wants, which were a part of the politeness theory introduced by Brown and

Levinson. More specifically, the aim was to determine whether positive and

negative politeness strategies would affect the quality of answers a chatbot

produces. The chatbot was asked 10 questions in total, grouped into five sets of

questions that prompted answers followed by argumentation. Each set contained

two questions on the same topic, one question asked utilizing positive politeness

strategies, and the other question asked using negative politeness strategies. The

results show that 1) there were no significant differences in the answers given to

the two types of questions, 2) the chatbot did not select only one type of

politeness strategies, 3) more elaborate answers to one of the questions from the

sets were produced in relation to the topic of that question rather than the

politeness strategies. The most dominant argumentation structure is based on

coordinative arguments, with causal schemes appearing most frequently. All of

the answers are acceptable in the sense that they give coherent arguments for a

standpoint, except for one answer where the opposite standpoints were provided.

Keywords: argumentation, politeness theory, positive and negative politeness

strategies, chatbot, artificial intelligence