This article presents two experimental studies investigating the impact of language errors in online dating profiles on impression formation. A first study examined whether language errors have a negative effect on perceptions of attraction and dating intention and whether this effect is moderated by the presence of visual information, that is, the profile picture. This 2 (Language Errors/No Language Errors) × 2 (Visible/Blurred Picture) experiment revealed that language errors negatively affect perceptions of social and romantic attraction and that a visible picture on a profile positively affects perceptions of physical attraction. Study 2 focused on mechanical, rule-based, and informal language errors, which can each be attributed to different personality traits. Mechanical and rule-based errors lead to lower scores on, respectively, perceived attentiveness and intelligence, which in turn lead to lower attraction and dating intention scores. These results highlight the importance of error-free language use as a cue for attractiveness.

This study uses two methods to examine whether online daters looking for a long-term relationship behave linguistically different in their profile texts compared to daters seeking casual relationships. To investigate these linguistic differences, 12,310 existing Dutch dating profiles were analyzed using the Linguistic Inquiry and Word Count (LIWC) program and a word-based classifier. Results of both methods suggest there are reliable differences in the linguistic behavior long-term and casual relationship seekers employ in their dating profiles: long-term relationship seekers mention more topics that are relevant when looking for a long-term relationship, such as internal personality traits and qualities. Additionally, long-term relationship seekers seem to self-disclose more in their profile texts by providing more personal information and using more I-references. Profile texts of casual relationship seekers are more diffuse and harder to classify. Moreover, the study demonstrates that using a multi-method approach, with LIWC and a data-driven word-based classifier, provides a deeper understanding of linguistic differences between the two relationship seeking groups.

This study investigates how online dating profiles, consisting of both pictures and texts, are visually processed, and how both components affect impression formation. The attractiveness of the profile picture was varied systematically, and texts either included language errors or not. By collecting eye tracking and perception data, we investigated whether picture attractiveness determines attention to the profile text and if the text plays a secondary role. Eye tracking results revealed that pictures are more likely to attract initial attention and that more attractive pictures receive more attention. Texts received attention regardless of the picture’s attractiveness. Moreover, perception data showed that both the pictorial and textual cues affect impression formation, but that they affect different dimensions of perceived attraction differently. Based on our results, a new multimodal information processing model is proposed, which suggests that pictures and texts are processed independently and lead to separate assessments of cue attractiveness before impression formation.

Psychologically motivated, lexicon-based text analysis methods such as LIWC (Pennebaker et al., 2015) have been criticized by computational linguists for their lack of adaptability, but they have not often been systematically compared with either human evaluations or machine learning approaches. The goal of the current study was to assess the effectiveness and predictive ability of LIWC on a relationship goal classification task. In this paper, we compared the outcomes of (1) LIWC, (2) machine learning, and (3) a human baseline. A newly collected corpus of online dating profile texts (a genre not explored before in the ACL anthology) was used, accompanied by the profile writers' self-selected relationship goal (long-term versus date). These three approaches were tested by comparing their performance on identifying both the intended relationship goal and content-related text labels. Results show that LIWC and machine learning models both correlate with humans in terms of content-related label assignment. Furthermore, LIWC's content-related labels corresponded more strongly to humans than those of the machine learning model. Moreover, all approaches were similarly accurate in predicting the relationship goal.

Language accommodation in online dating profiles: Effects of education level and type of dating site on language useThis study investigates whether online dating profile owners accommodate their language use based on the type of dating site used. Following Communication Accommodation Theory, we expect that highly educated online daters’ language use on a dating site designed explicitly for the highly educated is different from language use of highly educated people on a general dating site with users of both low and high education levels. Specifically, highly educated dating site users will accommodate their language use depending on the dating site they use. Based on an analysis of 1570 profile texts from two Dutch dating sites, we found that profiles of highly educated users on a general dating site differed from those written by highly educated daters on the high-education dating site on some but not all measures of lexical complexity and on the amount of language errors made. On the high-education dating site, the highly educated daters used some more linguistic elements that are associated with a high education level than those on the general site. This suggests that minor contextual differences in expected audience can already induce linguistic accommodation.