Segment transfer: a consequence of a dynamic system

Abstract

This article presents the foundations of the Feature Competition Model (FCM) of segment transfer. The FCM is a proposal to explain how L2 sounds are mapped on to L1 phonological categories. Like previous analyses on segment transfer, the FCM assumes that not all features are of the same prominence in a given phonemic inventory and that feature prominence can be determined through underspecification. Unlike previous analyses, the FCM adopts a dynamic approach to phonology, one which assumes that features do not have discrete values, rather ones which are continuous, of greater or lesser prominence in an inventory. A specific metric for calculating prominence is given, and hypotheses for three L1-L2 contexts are generated and tested. The results of an experiment suggest that the metric has predictive power, but that certain refinements of the formula are necessary. Finally, implications the FCM has for our understanding of developing L2 speech patterns are discussed.

Sink positive: Linguistic experience with th substitutions influences nonnative word recognition

“…1 Hancin-Bhatt (1994a, 1994b showed (using good listening conditions) that German L2 learners perceive English /θ/ 48% of the time as /θ/, 5% as /s/, 0% as /t/, and 22% as /f/. Despite methodological differences, both studies clearly show that /θ/ is perceptually most often confused with /f/ for both L2 and L1 listeners.…”

“…Several studies have not only described /θ/ substitutions across various L2 learners of English, but have also tried to determine the cross-linguistic differences in /θ/ production on the basis of the dissociation between perception and production of the English /θ/ (e.g., Brannen, 2002;Hancin-Bhatt, 1994a;Teasdale, 1997). Many phoneme identification and phoneme confusion studies have shown that /θ/ is perceptually most often confused with /f/ by native as well as by various nonnative listeners, and less frequently confused with /t/ or /s/ (e.g., Brannen, 2002;Cutler, Weber, Smits, & Cooper, 2004;Hancin-Bhatt, 1994a, 1994bMiller & Nicely, 1955;Tabain, 1998). For example, Cutler et al showed that under a 0-dB signal-to-noise ratio, Dutch L2 learners of English perceive English /θ/ 12.1% of the time as /θ/, 0.4% as /s/, 6.3% as /t/, and 13.3% as /f/ (for a comparison, American participants perceive /θ/ 18.3% of the time as /θ/, 0% as /s/, 5.4% as /t/, and 13.3% as /f/).…”

Phonetic adaptation in non-native spoken dialogue: Effects of priming and audience design

“…Many second language learners produce L2 pronunciations that are notably unlike those of native speakers, often showing a heavy influence of L1 sound inventory and sound patterning (e.g., Best, McRoberts, & Goodell, 2001;Brannen, 2002;Clements, 2001;Eckman & Iverson, 2013;Flege & Eefting, 1987;Hancin-Bhatt, 1994;Sirsa & Redford, 2013;White & Mattys, 2007). Our experimental design took advantage of the fact that ambiguities arise when non-native speakers fail to make necessary contrasts in the target language.…”

Beginning L2 production errors in ASL lexical phonology

“…In the L1 transfer model, the incorporation of the L2 lexical system is based on the constraints involved phonologically in the L1 phonology system (Flege 1981(Flege , 1991(Flege , 1992(Flege , 1995Ringbom 1978Ringbom , 1992Brown 1998Brown , 2000. The L2 learning of a segment involves the acquisition of representations of contrast between L2 and L1 segments, not just phonetic ability to produce the segment (Broselow 1984;Bley-Vroman 1989;Hancin-Bhatt 1994;Leather & James 1991. If segments in L1 and L2 differ, and if L2 learners substitute L1 for L2, they make L1 transfer errors (Broselow & Finer 1991;Hancin-Bhatt 1994;Gass 1996).…”