This paper investigates how fourteen fluency features interact with one another in the interlanguage of learners of English as compared to native speakers to form so-called fluency dimensions. The study is based on the time-aligned and annotated version of the French component of the Louvain International Database of Spoken English Interlanguage (Gilquin, De Cock, and Granger 2010), which is a collection of 50 interviews of Belgian French-speaking university students of English. The native speaker data comes from LINDSEI's native speaker counterpart, the Louvain Corpus of Native English Conversation (De Cock 2004).
"Learner (like native speaker) fluency is generally defined as an interplay between a range of features such as filled and unfilled pauses, reformulations, repetitions, or temporal measures (e.g. Götz 2013). To date, however, many studies have been restricted to the investigation of one of those features at a time (Bell 2003; Bosker et al. 2014). Moreover, few studies have considered the potential interaction(s) between fluency features or tried to identify underlying dimensions of fluency. One notable exception is the work by Skehan and Tavakoli, who identified three fluency dimensions, namely speed, breakdown, and repair fluency (Skehan 2003; Tavakoli and Skehan 2005; Tavakoli 2016).This paper investigates how fourteen fluency features interact with one another in the interlanguage of learners of English as compared to native speakers to form so-called fluency dimensions. The study is based on the time-aligned and annotated version of the French component of the Louvain International Database of Spoken English Interlanguage (Gilquin, De Cock, and Granger 2010), which is a collection of 50 interviews of Belgian French-speaking university students of English. The native speaker data comes from LINDSEI's native speaker counterpart, the Louvain Corpus of Native English Conversation (De Cock 2004).A Principal Components Analysis was used to identify fluency dimensions. While a speed and repair fluency dimension could indeed be identified in both learner and native speech, three or two new dimensions also emerged from the analysis (in learner and native speech respectively). A closer look at the association of features within each dimension further revealed that learner filled pauses, unlike native filled pauses, are not associated with unfilled pauses in the speed fluency dimension, but are grouped together with discourse markers in a separate dimension. These results shed a new light on Skehan and Tavakoli's studies."
