Alastair J. Gill

Language with Character: A Stratified Corpus Comparison of Individual Differences in E-Mail Communication.

To what extent does the wording and syntactic form of peoples writing reflect their personalities? Using a bottom-up stratified corpus comparison, rather than the top-down content analysis techniques that have been used before, we examine a corpus of e-mail messages elicited from individuals of known personality, as measured by the Eysenck Personality Questionnaire–Revised (S. Eysenck, Eysenck, & Barrett, 1985). This method allowed us to isolate linguistic features associated with different personality types, via both word and part-of-speech n-gram analysis. We investigated the extent to which extraversion is associated with linguistic features involving positivity, sociability, complexity, and implicitness and neuroticism is associated with negativity, self-concern, emphasis, and implicitness. Numerous interesting features were uncovered. For instance, higher levels of extraversion involved a preference for adjectives, whereas lower levels of neuroticism involved a preference for adverbs. However, neither positivity nor negativity was as prominent as expected, and there was little evidence for implicitness.

The language of emotion in short blog texts

Emotion is central to human interactions, and automatic detection could enhance our experience with technologies. We investigate the linguistic expression of fine-grained emotion in 50 and 200 word samples of real blog texts previously coded by expert and naive raters. Content analysis (LIWC) reveals angry authors use more affective language and negative affect words, and that joyful authors use more positive affect words. Additionally, a co-occurrence semantic space approach (LSA) was able to identify fear (which naive human emotion raters could not do). We relate our findings to human emotion perception and note potential computational applications.

Large scale personality classification of bloggers

Personality is a fundamental component of an individuals affective behavior. Previous work on personality classification has emerged from disparate sources: Varieties of algorithms and feature-selection across spoken and written data have made comparison difficult. Here, we use a large corpus of blogs to compare classification feature selection; we also use these results to identify characteristic language information relating to personality. Using Support Vector Machines, the best accuracies range from 84.36% (openness to experience) to 70.51% (neuroticism). To achieve these results, the best performing features were a combination of: (1) stemmed bigrams; (2) no exclusion of stopwords (i.e. common words); and (3) the boolean, presence or absence of features noted, rather than their rate of use. We take these findings to suggest that both the structure of the text and the presence of common words are important. We also note that a common dictionary of words used for content analysis (LIWC) performs less well in this classification task, which we propose is due to their conceptual breadth. To get a better sense of how personality is expressed in the blogs, we explore the best performing features and discuss how these can provide a deeper understanding of personality language behavior online.

In CMC we trust: the role of similarity

This paper examines how different forms of linguistic similarity in a text-chat environment relate to the establishment of interpersonal trust. Sixty-two pairs played an iterative social dilemma investment game and periodically communicated via Instant Messenger (IM). Novel automated and manual analysis techniques identify linguistic similarity at content, structural and stylistic levels. Results reveal that certain types of content (some positive emotion words, task-related words), structural (verb tense, phrasal entrainment), and stylistic (emoticons) similarity characterize high trusting pairs while other types of similarity (e.g., negative emotion words) characterize low trusting pairs. Contrary to previous literature, this suggests that not all similarity is good similarity.

Coordination in Conversation and Rapport

We investigate the role of increasing friendship in dialogue, and propose a first step towards a computational model of the role of long-term relationships in language use between humans and embodied conversational agents. Data came from a study of friends and strangers, who either could or could not see one another, and who were asked to give directions to one-another, three subsequent times. Analysis focused on differences in the use of dialogue acts and non-verbal behaviors, as well as co-occurrences of dialogue acts, eye gaze and head nods, and found a pattern of verbal and nonverbal behavior that differentiates the dialogue of friends from that of strangers, and differentiates early acquaintances from those who have worked together before. Based on these results, we present a model of deepening rapport which would enable an ECA to begin to model patterns of human relationships.

Linguistic mimicry and trust in text-based CMC

This study examines the relationship between linguistic mimicry and trust establishment in a text-chat environment. Twenty-six participant pairs engaged in a social dilemma investment game and chatted via Instant Messenger (IM) after every five rounds of investment. Results revealed that, within chat sessions, lexical mimicry (repetition of words or word phrases by both partners) was significantly higher for high-trusting pairs than for low-trusting pairs, but that lexical mimicry across chat sessions was significantly higher for low-trusting pairs than for high-trusting pairs. Theoretical and applied implications are discussed.

“Should Be Fun—Not!”: Incidence and Marking of Nonliteral Language in E-Mail

According to Kreuzs principle of inferability, speakers tend to employ nonliteral language when it can reasonably be perceived by their conversational partner. In a computer-mediated communicative setting, such as e-mail, this suggests that the e-mail writer might use discourse tools that facilitate comprehension on the part of the recipient. The present study examined rates of usage for various forms of nonliteral language in 210 e-mail messages written by young adults. In 94.30% of all e-mails there was at least one nonliteral statement, and participants used an average of 2.90 nonliteral statements per e-mail. Results showed that forms of nonliteral language that are typically deemed to be riskier, such as sarcasm, were used much less frequently than other less risky forms, such as hyperbole, and were marked with discourse markers more often. This indicates that e-mail authors are sensitive to the risky nature of nonliteral language use in e-mail, yet are savvy to the tools available to them in this communicative medium.

Verbal irony use in personal blogs

Blogs are a widely growing form of computer-mediated communication used to achieve various personal and professional communicative goals. In the present study, we examined previously posted entries from 71 regular bloggers. We examined the blogs for the use of five forms of verbal irony: hyperbole, understatement, rhetorical question, sarcasm and jocularity. In addition, topic and emotional valence of the ironic utterances were examined. Results showed that hyperbole and understatement were more frequently used than the other forms of ironic language. Discussion of hobbies and social outings was the most commonly occurring topic of ironic language, and bloggers used verbal irony to convey both positive and negative intent. The results of this study demonstrated that adult bloggers do use a variety of forms of verbal irony in their personal blogs, despite the potential risk of being misunderstood.

Privacy dictionary: A new resource for the automated content analysis of privacy

This article presents the privacy dictionary, a new linguistic resource for automated content analysis on privacy-related texts. To overcome the definitional challenges inherent in privacy research, the dictionary was informed by an inclusive set of relevant theoretical perspectives. Using methods from corpus linguistics, we constructed and validated eight dictionary categories on empirical material from a wide range of privacy-sensitive contexts. It was shown that the dictionary categories are able to measure unique linguistic patterns within privacy discussions. At a time when privacy considerations are increasing and online resources provide ever-growing quantities of textual data, the privacy dictionary can play a significant role not only for research in the social sciences but also in technology design and policymaking.

Privacy dictionary: a linguistic taxonomy of privacy for content analysis

Privacy is frequently a key concern relating to technology and central to HCI research, yet it is notoriously difficult to study in a naturalistic way. In this paper we describe and evaluate a dictionary of privacy designed for content analysis, derived using prototype theory and informed by traditional theoretical approaches to privacy. We evaluate our dictionary categories alongside privacy-related categories from an existing content analysis tool, LIWC, using verbal discussions of privacy issues from a variety of technology and non-technology contexts. We find that our privacy dictionary is better able to distinguish between privacy and non-privacy language, and is less context-dependent than LIWC. However, the more general LIWC categories are able to describe a greater amount of variation in our data. We discuss possible improvements to the privacy dictionary and note future work.