The effects of acoustic versus phonetic similarity in non-native vowel perception have been the focus of many second language (L2) speech perception models, examining how non-native sounds are perceived and assimilated by listeners. These models use perceptual discrimination tasks (e.g., AX, AXB) that may elicit different modes of perception depending on the memory load and linguistic experience required in each. This paper examined how native Australian-English (AusE) speakers perceived naturally-produced Dutch vowels in a 4IAX task, less commonly used but believed to elicit more continuous perception and bypass conscious linguistic processing. Participants listened to six Dutch vowel pairs spoken by varying speakers and chosen for their acoustic distance from AusE vowels. Results showed that /ɪ-i/ was the least accurately discriminated compared to other vowel pairs, confirming predictions that L1-L2 acoustic distance is a driving force in non-native speech perception and suggesting that linguistic experience may affect perception during the 4IAX task.
The Perceptual Assimilation Model (PAM-L2) and the Second Language Linguistic Perception model (L2LP) examine how L2 learners perceive non-native or L2 sounds as a function of the similarity between their first language (L1) and the non-native language. This similarity is operationalized on the basis of phonetic or acoustic distance between the two languages and is usually tested in discrimination tasks that require listeners to compare various sounds. In many perceptual discrimination tasks, listeners are tasked with responding whether two speech sounds are the same or different (AX) or whether one sound belongs to the category of two other sounds (AXB).
However, the results of AXB tasks have been shown to vary with inter-stimulus interval (ISI), lexical or linguistic knowledge, or synthetic versus natural stimuli. A less-used perceptual discrimination task that is thought to impose fewer demands on short-term memory and allow for more continuous (as opposed to categorical) perception is the 4-interval forced choice (4IAX) task. Listeners are asked to discriminate between two pairs of stimuli and choose which pair is different. Importantly, this task is believed to allow listeners to bypass their linguistic knowledge, such as the influence of language background or lexical access.
The current study addresses this issue by presenting native AusE listeners with naturally-produced, isolated Dutch vowels in a 4IAX task. Vowels produced by three different speakers were used to examine how listeners use speaker variability during categorical perception. Additionally, we were interested in probing how this distance in ‘non-linguistic’ (speaker identity) and ‘linguistic’ (vowel identity) information interacts during non-native discrimination tasks, since acoustic distance varies greatly between and within speakers of the same language.
Participants were 8 native Australian-English speakers who were presented with six Dutch vowel pairs spoken by three different speakers. The Dutch vowels were chosen for their acoustic distance from AusE vowels, with some pairs being more similar to AusE vowels and others being less similar. The participants were asked to perform a 4IAX task, where they had to choose which pair of vowels was different from the other two pairs.
The results showed that the /ɪ-i/ pair was the least accurately discriminated compared to the other vowel pairs. This suggests that the acoustic distance between the Dutch vowels and the AusE vowels played a significant role in the participants' ability to discriminate between the vowels. The results also suggest that linguistic experience may affect perception during the 4IAX task, as the participants' performance was influenced by their language background.
The study's findings have implications for our understanding of non-native speech perception and the role of acoustic distance in this process. The results suggest that acoustic distance is a driving force in non-native speech perception, and that linguistic experience may affect perception during the 4IAX task. The study's use of naturally-produced stimuli and the 4IAX task design also provide new insights into the processing of non-native speech sounds and the role of speaker variability in this process.
