Abstract
Perception and learning of second-language (L2) tones present challenges for learners, including those from tonal backgrounds. A particular challenge arises from tone contrasts that share similar pitch contours but differ in pitch height, for example, Cantonese rising and level tones form such a challenging pair.Previous studies are abundant on perceptual difficulties involving tone contrasts in L2, but also report a variety of linguistic and non-linguistic influential factors, including the nature of tones, previous experience with tones, task types and so on. Moreover, variations in tonal correlates (such as low or high pitch height) and mixed patterns in learners’ performance in diverse tasks all prompt a question on effects of input sequence on tone perception. However, prior research on the impact of presentation order on tone perception has yielded inconsistent results with some suggesting better perception when lower tones are presented before higher tones while others revealing a contradictory pattern. Under the L2LP model, speech perception is affected by both the first language (L1) and non-linguistic factors throughout three stages of learning. In align with this framework, we designed a series of experiments in attempt to explore the nature of cross-language perception and learning of Cantonese tones.
This study aims to use three experiments with interconnected goals to answer whether and how linguistic and nonlinguistic factors such as task variables play a role in perceiving and learning challenging tones. We start with an examination on if and how listeners’ perception of tones is affected by task variables such as the presentation order, then we focus on whether and how the distribution of input in training influence perceptual learning of Cantonese tones.
Specifically, Experiment 1 uses two tasks to investigate whether and how the discrimination of tone contrasts is affected by the presentation order (low-high or high-low). Experiment 1a investigates the effect of presentation order on discriminating rising, level ad falling tone contrasts by Cantonese, Mandarin and English listeners. Results indicate robust effects of presentation order favoring low-high direction mainly in rising and level tone contrasts but not in falling tone contrasts. Based on the low discrimination of rising tone contrasts even by Cantonese listeners, Experiment 1b examines whether listeners show categorical perception in rising tones using a resynthesized low-to-high tonal continuum. Results suggest evident categorical patterns by Cantonese listeners and also significant effect of presentation order favoring low-high direction. Experiment 1a and 1b confirm asymmetrical perception of tones as a function of presentation order, revealing the facilitative role of the low-high direction. The perceptual asymmetry varies across listeners’ first languages (L1s: Cantonese, Mandarin, English) and tone types (rising, level, falling).
Experiment 1 discovers directional asymmetry and perceptual preference towards low-high presentation order, providing empirical foundations to further examine the impact of these variables in tone training. Thus, Experiment 2 sets out to assess whether different orders within tone contrasts (level vs rising) affect perceptual learning by Mandarin listeners under different training methods (low-high vs high-low distribution). Results indicate significantly better learning outcomes by participants trained with low-high ordered tone pairs in both level and rising tones. Generalization results also reveal stronger learning performances in novel stimuli of rising tones.
Given the evidence from training with single-talker stimuli, low variability (LV) input, in Experiment 2, Experiment 3 moves forward to test training effects of multi-talker materials, high variability (HV) input, and its interaction with presentation order. Results confirm significantly better performance during training among listeners trained with stimuli presented in low-high order, but the disparity between two training groups (low-high and high-low) decreases compared with Experiment 2. The LV input leads to limited improvement in learning level tones, whereas the HV input leads to notable improvement for both rising and level tones.
The findings from the whole study outline the developmental trend in perceiving and learning Cantonese tones. Learners’ better discrimination of the level tone contrasts before training and greater learning performances in rising tone contrasts after training together promote the hypothesis under the L2LP that speech learning entails re-ranking of perceptual correlates that can be gauged by both linguistic and non-linguistic factors. The perceptual advantages from the “low-high” order and the “high-variability” input provide practical implications to second language speech learning and to speech intervention for patients with speech disorders.
| Date of Award | 11 Sept 2024 |
|---|---|
| Original language | English |
| Awarding Institution |
|
| Supervisor | Bin LI (Supervisor) |