Up in the Air : a Linguistic and Computational Analysis of Alitalia in-Flight Magazine

Bilingual in-flight magazines provide air travellers with “infotainment”, but also represent a primary instrument to build “brand awareness” for the carrier, selling products and strengthening relationship with their customers. Our study is aimed at investigating the discourse features of Alitalia bilingual in-flight magazine, Ulisse, distributed on all routes throughout the network – in VIP lounges, at the main domestic and international airports, and in leading hotel chains. In order to carry out our analysis, we collected the English version of texts from 6 monthly issues, published between May 2015 and October 2016. The reach of in-flight magazines is interestingly broad, both from a qualitative and quantitative point of view. Qualitatively, the target readers are all travellers, from different countries, who find their personal complimentary copy in their seat pocket. Quantitatively, an average of 2.1 million passengers fly with Alitalia every month, with readership of 180,000 for each issue (http://www.mediaplaned.it/mediakiting.pdf). However, despite the importance of such genre, few academic studies have examined this phenomenon (Thurlow / Jaworky 2003, Maci 2012, Conradie 2013). Our work will be organized around the following research questions: 1) How are first person pronouns and corresponding determiners used in these texts? 2) How and to what extent are evaluative expressions used as manifestations to establishing identity? 3) Does the dataset show the presence of other devices employed by the writers to establish identity? And to what extent, do the text producers project cultural and ideological values, and whose values might these be? 4) What is/are the common sentiment(s) elicited by the texts and what are the most reccurent words? A combined framework of theoretical approaches and perspectives is used in this study in order to analyze the linguistic and multimodal (text classification and sentiment analysis) modes of communication in the text. More specifically, a qualitative Critical Discourse Analysis (Fairclough 2003, 2005; Wodak and Meyer 2001) approach was taken to examine textual elements, while a Discourse Analysis in the area of text evaluation (Hunston and Thompson 2001) was also brought together to provide an understanding of the meanings and social significance of texts. Finally, we decided to detect feelings emerging from the contents, using sentiment analysis tools and some procedures for text classification. The tools derive from Artificial Intelligence (A.I.). Thanks to cloud-based tools and procedures based on Natural Language Processing A.I. algorithms (Nasukawa and Yi, 2003; Lohr, 2012), it is possible to extract sentiments from content and from the web and analyze them just as any human being would do. The procedure is however much faster and allows to obtain more consistent results, considering that usually individuals, who should work together to analyze content, hardly come to an agreement on the sentiments elicited by texts.