Several nonnative speech perception models address the systematic phonetic relations between native and nonnative speech categories. The objective of this study is to examine the relative difficulty encountered by L2 learners in perceiving nonnative contrasts in high- and low-frequency words. The findings are discussed in relation to the NLM model.
This study presents the findings of two perception tasks that were conducted to examine the relative difficulty encountered by Cypriot Greek speakers of English in perceiving L2 sounds in high- and low-frequency words. The tasks involved a phoneme identification task (in line with Baker, 2006) and a minimal pair task (in line with Kkese, 2016). These tasks were preferred since they eliminate any semantic information from the context. The aim was to examine voicing and vowel length according to word familiarity. The participants involved 130 university students. Results are discussed in relation to the Native Language Magnet Model (NLM; Kuhl, 2000a, b, 1995, 1993; Iverson & Kuhl, 1996; 1995). The theory essentially claims that the listeners’ perceptual sound systems are composed of prototypes or best exemplars for each sound category. Prototypes act as magnets in L2 acquisition with reference to other sounds resulting in perceptual difficulties in the cases in which the L2 phones resemble a prototype. Therefore, nonnative contrasts may be difficult to discriminate when the prototype of an L1 category closely resembles the two L2 phones (Kuhl, 1993). This study examines whether the findings from the tasks fulfil the above predictions and specifically whether L2 listeners pay attention only to relevant L1 acoustic cues. Overall, the findings indicate task effect and difficulties with word familiarity since the item analysis suggested that low-frequency words are more difficult to perceive, especially with respect to consonants, in the phoneme identification task. The participants performed slightly better when dealing with vowels compared to consonants. Concerning the latter, the most difficult consonants were plosive, fricative, and nasal. Age, gender, years of L2 instruction and use, as well as word length were statistically significant factors for speech perception. The study then discusses some of the implications of the findings on the model.
