The effects of speaking task on L2 fluency
(2017) 4rth Learner Corpus Research Conference (LCR2017) — Location: Bolzano (Italy) (4.October.2017)
Files
No attached file found for this publication.
Details
- Authors
- Abstract
- Learner (like native speaker) fluency is characterised by an interplay between speed, breakdown and repair phenomena (e.g. De Jong 2016; Skehan 2003; Tavakoli & Skehan 2005). These include, among others, filled and unfilled pauses, reformulations, repetitions and temporal features such as speech rate. Research has provided valuable insights into the way these features interact with one another or are affected by factors such as mother-tongue background, time spent in an L2 environment or proficiency level (e.g. Freed 1995; Ginther, Dimova & Yang 2010; Götz 2013; Guz 2015; Towell, Hawkins & Bazergui 1996). With the advent of larger and more varied spoken corpora, a growing number of studies has turned to investigate how fluency might also be impacted by speaking task features such as task type (narrative, read-aloud task etc.), planning or level of interaction. Overall, research findings have revealed that variations in fluency measures can indeed partly be attributed to task type (e.g. Cucchiarini, Strik & Boves 2002; Foster & Skehan 1996) and that planning can have beneficial effects on learners’ fluency performance (e.g. Crookes 1989; Foster & Skehan 1999; 2009; Mehnert 1998). With respect to the level of interaction, research indicates that speakers seem to be more fluent in dialogues than in monologues in terms of speed, length of silent pauses and repair measures (Tavakoli 2016; see also Ejzenberg 2000). Against this backdrop, this study is an attempt to provide a more in-depth understanding into the extent to which speaking task influences the production of fluency features of French-speaking learners of English, and of B2 vs. C1 learners more particularly. For this purpose, filled and unfilled pauses, restarts, as well as speech rate and mean length of runs – which are all mentioned in the CEFR fluency descriptors (Council of Europe 2001) – are examined across three speaking tasks. The study is based on the French component of the Louvain International Database of Spoken English Interlanguage (Gilquin, De Cock & Granger 2010). Like the other components, LINDSEI-FR contains interviews of 50 intermediate to advanced French-speaking learners of English as a foreign language. Each interview is made up of three speaking tasks. First, a set topic is presented to the learner, who has some time to prepare what he/she is going to say. This task is followed by a spontaneous free discussion on various topics and, finally, by a monologic picture description task. These speaking tasks differ along a number of variables, including the degree of elicitation, of preparedness, as well as interactivity, which, based on previous research, are all likely to impact on the learner’s fluency. LINDSEI-FR interviews have been time-aligned at the level of words – a procedure that allows for the precise measurement of temporal variables such as speech rate or length of pauses – and subsequently semi-automatically annotated for a wide range of (dis)fluency features, including, among others, filled and unfilled pauses, restarts, repetitions, and discourse markers. Moreover, the fluency of each LINDSEI-FR learner has been assessed by three professionally-trained raters according to the CEFR grid and descriptors (inter-rater agreement reached an acceptable level). The grades were pooled and mean ratings computed per speaker. Results so far indicate that the frequency of unfilled pauses, speech rate as well as the mean length of runs are particularly affected by task properties, while the rate of filled pauses and restarts does not differ significantly across tasks. The monologic picture description task is, for example, characterized by a higher frequency of unfilled pauses and a lower speech rate as compared to the set topic and the free discussion task. The pre-planned set topic task differs from the spontaneous free discussion with respect to the mean length of runs. These result thus partly corroborate previous findings. Preliminary investigations into the fluency of B2 (n=22) and C1 (n=26) learners indicate that, although higher CEFR fluency level leads to better performances in the three tasks (i.e. fewer pauses and restarts, faster speech rate and longer speech runs), the previously mentioned differences between the tasks remain. Comparable analyses based on native speaker data from LOCNEC (the Louvain Corpus of Native English Conversation; De Cock 2004), LINDSEI’s native speaker counterpart, indicate that speaking task characteristics also influence the fluency of native speakers, though not to the same extent as for the learners.
- Affiliations
Citations
Dumont, A. (2017). The effects of speaking task on L2 fluency. 4rth Learner Corpus Research Conference (LCR2017), Bolzano (Italy). https://hdl.handle.net/2078.5/175735