Phonological prediction in spoken language processing

Sun, Yan

Permalink

https://hdl.handle.net/2142/117587 Copy

Description

Title

Phonological prediction in spoken language processing

Author(s)

Sun, Yan

Issue Date

2022-12-02

Director of Research (if dissertation) or Advisor (if thesis)

• Shih, Chilin
• Green, Jeffrey

Doctoral Committee Chair(s)

Shih, Chilin

Committee Member(s)

• Dell, Gary
• Christianson, Kiel
• Shosted, Ryan

Department of Study

Linguistics

Discipline

Linguistics

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

• phonological form prediction
• anticipatory tonal variation
• spoken-word recognition
• eye-tracking
• ERP

Abstract

The notion that prediction plays an important role in language comprehension has received growing acknowl- edgment in contemporary theories of language processing. One of the key and most widely cited pieces of empirical evidence for linguistic prediction comes from the landmark study by DeLong, Urbach, and Kutas (2005), who found evidence that readers can pre-activate the phonological form of predictable words in real-time sentence processing. However, the failure to replicate the key findings of DeLong et al. (2005) in some recent studies (e.g., Ito, Martin, & Nieuwland, 2017a; Nieuwland et al., 2018) has cast doubt on the robustness of such evidence. This dissertation re-evaluates phonological form pre-activation in predictive language processing using a novel experimental design. The main objectives of the current dissertation are to examine the causes of the divergent findings on this topic in the literature and to verify that modulations of event-related brain potentials (ERP)—such as the N400—at target words can also index lexical pre-activation. To achieve these goals, three studies are conducted in this dissertation. Study 1 uses a production experiment to investigate how the F0 realization of full tones in Standard Mandarin varies in anticipation of different following tones in connected speech. Two main types of anticipatory tonal variation have been attested: third-tone sandhi (3TS) and anticipatory tonal coarticulation (ATC). Study 2 uses two eye-tracking experiments to examine whether 3TS and ATC cues in speech signals can facilitate spoken-word recognition and whether listeners can employ such cues to make a tonal prediction of the next word. The results suggest a facilitatory effect of 3TS cues but not ATC cues on spoken-word recognition. Furthermore, 3TS cues are found to be used by listeners as a more reliable indicator of the tone of the following word than ATC cues. Study 3 uses an ERP experiment with a novel design to re-evaluate whether the phonological form of predictable words can be pre-activated in spoken language processing. ERP evidence for the benefits of confirmed predictions and for the costs of disconfirmed predictions are detected at target words when pre-target words contain 3TS cues, but not so when they contain ATC cues. Taken together, the findings of this dissertation contribute to resolving the dispute over phonological form pre-activation by showing that the lack of ERP effects at pre-target words does not necessarily deny lexical pre-activation; rather, it could indicate that the cues presented in the pre-target words are not a reliable indicator of the phonological form of the target words. This clarification is important for theories of predictive language processing. In addition, this dissertation also provides empirical evidence to support the prediction account of the N400 ERP component and thus supports the view that prediction is an important mechanism of language comprehension.

Graduation Semester

2022-12

Type of Resource

Thesis

Copyright and License Information

Copyright 2022 Yan Sun

Owning Collections