Perceptual flexibility in native and non-native speech perception and sentence processing: listener’s attention shifting across speech units and attention weight on cues

Ren, Xiaomu (2025) Perceptual flexibility in native and non-native speech perception and sentence processing: listener’s attention shifting across speech units and attention weight on cues. PhD thesis, University of Glasgow.

Abstract

Two overarching questions within the realm of speech perception and comprehension have been asked: Why is perceiving native speech effortless, while processing non-native speech is challenging? In this thesis, I address these questions by exploring how native and non-native listeners differ in their perceptual flexibility. I apply the concept of perceptual flexibility as a cognitive mechanism of listeners to shed light on the potential differences demonstrated by native and non-native listeners. This mechanism refers to how listeners adapt to variations in acoustic signals during speech perception and comprehension. Listeners with high perceptual flexibility are predicted to actively shift attention across and detect speech units of different sizes during real-time speech unit perception, and to assign attention weights to and flexibly integrate multiple cues during real-time sentence processing. In contrast, listeners with reduced perceptual flexibility are predicted to be less able to actively shift attention across units of different sizes and to rely consistently on certain cues with a reduced ability to integrate multiple cues. Therefore, I pose a general question: Do non-native listeners exhibit comparable perceptual flexibility to native listeners in speech unit perception and sentence processing? This thesis employs monitoring and visual-world eye-tracking paradigms to test listeners’ perceptual flexibility, which is manifested in both low-level speech perception detecting perceptual units of different sizes—and high-level sentence processing—utilizing and integrating multiple cues.

Experiments 1 and 2 are phoneme and syllable monitoring experiments using English (Experiment 1) and Mandarin (Experiment 2) pseudo-words. In each experiment there was a native listener group (English native speakers in Experiment 1, Mandarin native speakers in Experiment 2) and a non-native group (Mandarin native speakers learning English as L2 in Experiment 1, English native speakers learning Mandarin as L2 in Experiment 2). To observe how listeners flexibly shift their attention between phonemes and syllables, I manipulated the stimuli in two ways. First, I created stimuli with an artificial accent, which served as bottom-up information. Second, I used prior knowledge of the artificial accent as top-down information. I then examined how reaction times (RTs) to phoneme and syllable targets changed under these manipulations. The results revealed that both English and Mandarin listeners exhibited comparable perceptual flexibility in detecting units of different sizes in the English context (Experiment 1), with greater sensitivity to syllables than to phonemes. In the Mandarin context (Experiment 2), English listeners showed higher perceptual flexibility than Mandarin listeners. These findings shed light on how perceptual units are represented and retrieved during native and non-native speech unit perception, providing experimental support that syllable processing is more easily disrupted than phoneme processing, especially by distorted bottom-up input, suggesting that phonemes may be a more robust unit than syllables during speech perception.

While Experiments 1 and 2 examined perceptual flexibility in perceiving sublexical units, Experiments 3 and 4 looked at sentence-level processing. Experiments 3 and 4 employed a visual-world eye-tracking paradigm with English (Experiment 3) and Mandarin (Experiment 4) sentence stimuli, and utilized native and non-native listener groups as for Experiments 1 and 2. Experiments 3 and 4 investigated how native and non-native listeners adopt different perceptual strategies during real-time sentence processing, focusing on how they allocate attention to various cues and integrate them flexibly. This eye-tracking experiment particularly focused on prosodic cues, verb semantics, and whether information is repeated or new in a broader discourse context. Results showed a complex three-way interaction among L1, prosody, and verb semantics. With both English and Mandarin stimuli, when processing sentences containing old information, non-native listeners tended to adopt perceptual strategies similar to those of native listeners. Specifically, both native and non-native listeners in both languages exhibited effect of both semantics and prosody, with the semantic effect being exaggerated when the target carried a prosodic accent. However, when processing sentences containing new information, non-native and native listeners exhibited divergent patterns, the specifics of which depended on whether the stimulus language was English or Mandarin. Native listeners can interactively combine both semantic and prosodic information across a variety of contexts. Non-native listeners can also do this, but only with familiar repeated information. In contexts when they must handle new information, their ability to integrate different cues is reduced.

Thus, the main findings of the thesis are threefold. First, in the perception of low-level sublexical speech units, native and non-native listeners generally demonstrated comparable levels of perceptual flexibility when shifting their focus flexibly and actively between phonemes and syllables, with syllable processing being more susceptible to disruption than phoneme processing. Second, in high-level sentence processing, native and non-native listeners began to show a divergence in their ability to utilize and integrate multiple cues, particularly when processing sentences containing new information. Third, during high-level sentence processing, perceptual flexibility was influenced not only by a listener’s linguistic experience but also by the characteristics of the language they were processing.

• Item Type: Thesis (PhD)
• Qualification Level: Doctoral
• Subjects: P Language and Literature > P Philology. Linguistics
• Colleges/Schools: College of Arts & Humanities > School of Critical Studies > English Language and Linguistics
• Supervisor's Name: Cohen, Dr. Clara and Smith, Dr. Rachel
• Date of Award: 2025
• Depositing User: Theses Team
• Unique ID: glathesis:2025-85087
• Copyright: Copyright of this thesis is held by the author.
• Date Deposited: 23 Apr 2025 14:14
• Last Modified: 22 Aug 2025 07:46
• Thesis DOI: 10.5525/gla.thesis.85087
• URI: https://theses.gla.ac.uk/id/eprint/85087