“…Approaches that base their analyses on spontaneous speech data have often been criticized for using anecdotal evidence. To counter such problems, there have been a number of laboratory studies into the functional use of facial expressions, either by eliciting such expressions from real speakers in specifically designed linguistic or social contexts (Dohen, 2005;Dohen & Loevenbruck, 2009;Dohen, Loevenbruck, Cathiard, & Schwartz, 2004;Krahmer & Swerts, 2007;Scarborough et al, 2009;, or by using an analysis-by-synthesis procedure whereby the effect of specific variations in the face on observers are tested through controlled manipulations of synthetic talking heads (Cassell, Vihjá lmsson, & Bickmore, 2001;Granström, House, & Lundeberg, 1999;Krahmer, Ruttkay, Swerts, & Wesselink, 2002;Krahmer & Swerts, 2004;Pelachaud, Badler, & Steedman, 1996). Such studies showed that facial expressions are indeed functionally relevant as they are exploited to support a broad range of communicative functions (e.g., to highlight important information, or to reveal attitudinal or emotional correlates of speaker utterances).…”