Visual cues during interaction: Are recasts different from noncorrective repetition?

Abstract

Visual cues may help second language (L2) speakers perceive interactional feedback and reformulate their nontarget forms, particularly when paired with recasts, as recasts can be difficult to perceive as corrective. This study explores whether recasts have a visual signature and whether raters can perceive a recast’s corrective function. Transcripts of conversations between a bilingual French–English interlocutor and L2 English university students ( n = 24) were analysed for recasts and noncorrective repetitions with rising and declarative intonation. Videos of those excerpts ( k = 96) were then analysed for the interlocutor’s provision of visual cues during the recast and repetition turns, including eye gaze duration, nods, blinks, and other facial expressions (frowns, eyebrow raises). The videos were rated by 96 undergraduate university students who were randomly assigned to three viewing conditions: clear voice/clear face, clear voice/blurred face, or distorted voice/clear face. Using a 100-millimeter scale with two anchor points (0% = he’s making a comment, and 100% = he’s correcting an error), they rated the corrective function of the interlocutors’ responses while their eye gaze was tracked. Raters reliably distinguished recasts from repetitions through their ratings (although they were generally low), but not through their eye gaze behaviors.

The visual signature of non-understanding: A systematic replication of McDonough, Trofimovich, Lu, and Abashidze (2019)

“…Carpenter, Jeon, MacGregor, and Mackey (2006) reported that external observers of recasts and non-corrective repetitions did not mention visual cues as being useful for differentiating between episode types. Similarly, McDonough et al (2020) found that external raters either associated the same visual cue with both recasting and non-corrective repetition or claimed that a visual cue was unique to a conversational move when it actually occurred in both episode types. Even though they could not identify the visual cues associated with each episode type, those raters did attribute significantly higher ratings of corrective intent to recasts than to non-corrective repetitions.…”

“…Other visual cues such as nodding, head shakes, and pointing have been used by second language (L2) instructors when providing students with corrective feedback (Davies, 2006; Faraco & Kida, 2008; Wang & Loewen, 2016). For instance, during dyadic interaction with L2 English speakers, a listener provided more head nods and blinks prior to recasting but longer eye gaze with more facial expression after non-corrective repetition (McDonough, Trofimovich, Lu, & Abashidze, 2020).…”

Exploring the depths of second language processing with eye tracking: An introduction

“…As demonstrated in this special issue, eye tracking, an exquisitely versatile methodology, is an excellent tool to do this because it can show how language users allocate, shift, or divide their attention between multiple channels. Thus, the present special issue features eye tracking used in spoken discourse processing (McDonough et al, 2020), reading-while-listening (Conklin et al, 2020), the visual world paradigm (Andringa, 2020), and in an integrated writing task (Michel et al, 2020). Although the specific research questions in each study differ, the authors share a common interest in how language users integrate multiple sources of information (e.g.…”

“…verbal and nonverbal; spoken and written; words and images) into one coherent, multimodal representation that will help the participants accomplish their goals. For instance, McDonough et al (2020), working in a social paradigm, explored whether there is a visual signature to the delivery of corrective feedback during oral interaction. The researchers measured a third party’s eye gaze to the speaker’s face as he or she (the observer of the conversation) was asked to rate whether the speaker’s utterance was meant as a simple (non-corrective) repetition of a previous utterance or a (corrective) recast.…”

Examining Rater Perception of Holds as a Visual Cue of Listener Nonunderstanding

“…By identifying the nonverbal signals of repair practiced by multiple speakers of different languages in diverse conversational settings, these researchers have demonstrated generality in repair practices, which can be understood as the extent to which practices are organized in the same way across contexts (see Chenail, 2010, for discussion of generalizability and related constructs in qualitative research). Inspired by this line of research, we were also interested in generality and carried out a series of studies that examined whether nonverbal aspects of repair practices, specifically clarification requests (McDonough et al, 2019, 2021) and recasts (McDonough et al, 2015, 2020a, 2020b), were organized similarly in conversations between university students.…”