The Telling Tail: Signals of Success in Electronic Negotiation Texts

The Telling Tail: Signals of Success in Electronic Negotiation Texts Marina Sokolova IRO, Universit´e de Montr´eal Montr´eal, Qu´ebec, Canada sokolovm@iro.umontreal.ca Vivi Nastase EML Research, gGmbH Heidelberg,Germany nastase@eml-research.de Stan Szpakowicz SITE, University of Ottawa Ottawa, Ontario, Canada ICS, Polish Academy of Sciences Warsaw, Poland szpak@site.uottawa.ca Abstract We analyze the linguistic behaviour of par- ticipants in bilateral electronic negotiations, and discover that particular language char- acteristics are in contrast with face-to-face negotiations. Language patterns in the later part of electronic negotiation are highly in- dicative of the successful or unsuccessful outcome of the process, whereas in face-to- face negotiations, the first part of the nego- tiation is more useful for predicting the out- come. We formulate our problem in terms of text classification on negotiation segments of different sizes. The data are represented by a variety of linguistic features that cap- ture the gist of the discussion: negotiation- or strategy-related words. We show that, as we consider ever smaller final segments of a negotiation transcript, the negotiation- related words become more indicative of the negotiation outcome, and give predictions with higher Accuracy than larger segments from the beginning of the process. 1 Introduction We use language every day to convince, explain, ma- nipulate and thus reach our goals. This aspect of language use is even more obvious in the context of negotiations. The parties must reach an agree- ment on the partitioning or sharing of a resource, while each party usually wants to leave the negotia- tion table with the larger piece of the pie. These ten- dencies become stronger when negotiators use only electronic means to communicate, that is to say, par- ticipate in electronic negotiations. In face-to-face contact, prosody and body language often have a crucial role in conveying attitudes and feelings. E- negotiators, on the other hand, must rely only on texts. We perform automatic analysis of the textual data in e-negotiations. We identify linguistic expres- sions of such negotiation-specific behaviour that are indicative of the final outcome of the process – suc- cess or failure – and observe how powerful a tool language is in helping people get what they want. In this paper we focus on the negotiation as an on- going process. We analyze the linguistic features of messages exchanged at various points in the course of the negotiation, to determine the time frame in which the outcome becomes decided. From our ex- perimental point of view, we determine the segment of the negotiation which is most predictive of the outcome. There is an imposed three-week deadline in the electronic negotiations that we analyze. We hypothesize that the pressure of the deadline is re- flected in the messages exchanged. The messages written later in the process are more indicative of the outcome of the process. Our empirical results support this hypothesis; an analysis of the linguis- tic features that make this prediction possible shows what the negotiators’ main concerns are as the dead- line draws near. Here is what our results contribute to the field of …

Content truncated to 500 words