Untangling Linguistic Salience

The concept of linguistic salience is broadly used within sociolinguistics to account for processes as diverse as language change (Kerswill & Williams, 2002) and language acquisition (Ellis, 2016) in that salient forms are e.g. more likely to undergo change, or are often acquired earlier than other features. However, the meaning of salience is “notoriously difficult to quantify” (Hickey, 2000, p. 57) and definitions of the term given in the literature often differ to such a degree that one could dispute whether the concept of salience has explanatory value (cf. Rácz, 2013). Accordingly, what makes a particular linguistic feature salient is contested: some argue that salience can be defined by linguistic traits such as loudness, high word-frequency, or a greater articulatory effort, whereas others argue that salience is a result of associations with social factors (cf. Kerswill & Williams, 2002). In a pilot study, we used eye-tracking to collect pupil dilation data while participants listened to spoken samples that were hypothesized to be either salient or not. These differences in salience were based on notions taken from the literature and included traits such as acoustical prominence, gender violations, loudness and differing realizations of the consonants /r/ and /v/. We were able to show that pupil size increased significantly for salient variables in the categories acoustic prominence, gender and loudness, pointing towards an increase in brain activity for these variables. In this poster, we propose to untangle how the concept is used. To those ends, we conducted a review of the literature on salience and identified different ways of operationalizing it. We conclude by discussing how salience could be decomposed in terms of other notions such as frequency, surprisal and markedness. We then propose a series of experiments using eye-tracking and ERP experiments