Author: Zhang, Changwei
Title: The perception and production development of Cantonese syllable-final segments by Mandarin speakers
Advisors: Leung, Man-tak (CBS)
Chen, Si (CBS)
Degree: Ph.D.
Year: 2022
Subject: Second language acquisition
Cantonese dialects -- Phonetics
Cantonese dialects -- Study and teaching -- Chinese speakers
Hong Kong Polytechnic University -- Dissertations
Department: Department of Chinese and Bilingual Studies
Pages: xvii, 213 pages : illustrations
Language: English
Abstract: Previous research on the second language (L2) phonetic learning stresses the influence of learners' native language (L1) experience and the modulation of perception-production interface. In this dissertation, Mandarin speakers' learning of Cantonese syllable-final segments ([-p], [-t], [-k], [-m], [-n], [-ŋ] and [-Ø]) is taken as a starting point to disentangle the research questions centering around these two issues.
Under the influence of L1 experience, L2 learners are apt to link L2 speech sounds to those previously existing L1 categories when confronting an L2. Several predominant L2 learning models including the Speech Learning Model (Flege, 1995), the Second Language Linguistic Perception Model (Escudero, 2005), the Perceptual Assimilation Model to Second Language (Best & Tyler, 2007) and the revised Speech Learning Model (Flege & Bohn, 2021) consistently prioritize the important role of cross-L1-L2 similarity in predicting the learning outcome and learning difficulty of L2 speech learning. However, the definitions and measurement methods of the cross-L1-L2 similarity are not always in agreement, which could lead to a distinct interpretation of these theories.
Researchers come up with divided opinions about how perception would interact with production during the development of L2 phonetic learning. To date, the production-precedence pattern (Goto, 1971; Sheldon & Strange, 1982; McCandliss, Fiez, Protopapas, Conway, & McClelland, 2002; Hattori & Iverson, 2009; Ingvalson, McClelland, & Holt, 2011), the perception-precedence pattern (Flege, 1995), and the perception-production co-evolve pattern (Flege & Bohn, 2021) have been proposed. The relationship between perception and production remains debatable and it could lead to different decisions on training methods of L2 speech learning. However, the above studies seldom specify whether perception would interact diversely with production according to different targets. As remarked by Cheng & Zhang (2009), the perception-production relationship varies depending on target segments. It could be possible that some L2 targets achieve correct perception before successful pronunciation, while some L2 targets might take the alternative path to develop production before perception or to co-evolve perception and production together during the developmental procedure. If so, intervention should be appropriately administered to better facilitate L2 learners' development of different targets.
The research questions were fourfold: (1) to measure the cross-language perceptual similarity of syllable-final segments between Mandarin and Cantonese by Mandarin speakers; (2) to interpret the development of Cantonese syllable-final segments by Mandarin speakers in the framework of SLM-r. Then, a general discussion would supplement the SLM-r predictions of the current study with alternative PAM-L2 and L2LP hypotheses and advance heuristic amendments to the application of aforementioned L2 models; (3) to delineate the perception-production interface during the learning procedure of targets; and (4) to testify the effectiveness of perception-only training on promoting Mandarin speakers' learning of targets.
Four experiments were incorporated to disentangle these research questions. As revealed by the results of the Perceptual Assimilation Test (Experiment 1) in Research Question (1), Cantonese [-Ø] represented a reasonably good exemplar to Mandarin [-Ø]. Moderately fit exemplars were represented by Cantonese [-p], [-t] and [-k] to Mandarin [-Ø], Cantonese [-m] to Mandarin [-ŋ], Cantonese [-n] to Mandarin [-n] and Cantonese [-ŋ] to Mandarin [-ŋ]. Accordingly, Cantonese [-Ø] was classified as an identical category with the least learning difficulties and the other six targets were categorized as less learnable similar sounds to their Mandarin counterparts in SLM-r.
These SLM-r hypotheses were attested in Research Question (2) with the developmental patterns of Cantonese syllable-final segments by an experimental group of Mandarin speakers through the pre-test (Experiment 2), seven sessions of perception training (Experiment 3), and the post-test (Experiment 4). In line with the predictions, identical target [-Ø] generally outperformed similar targets [-p], [-t], [-k], [-m], [-n] and [-ŋ] in the pre-test. During the intervention, learnable identical target ([-Ø]) was developed earlier and with better performances than those more difficult similar sounds ([-m], [-n], [-ŋ], [-p], [-t] and [-k]) by Mandarin speakers.
The SLM-r hypotheses, however, failed to clarify why Cantonese [-p], [-t], [-k], [-m], [-n] and [-ŋ] were all classified as similar sounds but targets [-t] and [-k] were more difficult than other similar targets for Mandarin speakers. As both Cantonese [-t] and [-k] were mapped to Mandarin [-Ø] without significant differences in the similarity rating scores, Cantonese contrast [-t]-[-k] can be categorized as Single-category Assimilation in PAM-L2 (Best & Tyler, 2007) or the learning of NEW scenario in L2LP (Escudero, 2005). It is predicted that both cases are the most difficult for Mandarin speakers. Such PAM-L2 and L2LP hypotheses provide a possible account for the least satisfying performances and learning outcomes of targets [-t] and [-k] by Mandarin speakers in the present study.
Other than the influence of L1 experience on L2 phonetic learning as proposed by the abovementioned theories, it is also possible that the suspension of contrastive distribution of coronal and dorsal targets in Cantonese could pose more learning difficulty in Mandarin speakers' development of targets [-t] and [-k]. Cantonese participants' performances in Experiment 2 (the pre-test) generally support the coronal-dorsal merger and the alveolarization merging tendency as observed by previous studies (see Bauer, 1979; Yeung, 1981; Chen, 1999; Zee, 1999a & 1999b; Law, Fung, & Bauer, 2001; Wong, 2005; Ding, 2010; Bauer & Benedict, 2011; To, Cheung, & McLeod, 2013; To, McLeod, & Cheung, 2015). The relatively lower learnability of these two targets can result from the immersion of the phonetic variants of [-t]-[-k] merger in daily Cantonese by Mandarin speakers. Extant studies of theoretical models in L2 phonetic learning all target contrastively distributed L2 sounds with little attention paid to a situation where target L2 sounds might be experiencing a merger. The coronal-dorsal merging tendency of Cantonese syllable-final segments observed in this study opens up a chance to unveil this issue. Therefore, heuristic amendments to the application of theoretical models in L2 phonetic learning are advanced. It is suggested that a more comprehensive prediction of L2 phonetic learning should integrate the perceived cross-L1-L2 similarity with the distribution patterns of L2 targets.
As concerned by Research Question (3), diverse patterns of perception-production interface as a function of the learnability of different targets in the L2 phonetic learning is pointed out. In the case of less learnable similar targets [-p], [-t], [-k], [-m], [-n] and [-ŋ], the perception was positively correlated with production and performed a perception-precedence pattern as proposed by Flege (1995). The successful perception of these similar targets was developed before their production. The case of the identical target [-Ø] failed to provide overt support to any of the production-precedence pattern (Goto, 1971; Sheldon & Strange, 1982; McCandliss et al., 2002; Hattori & Iverson, 2009; Ingvalson et al., 2011), the perception-precedence pattern (Flege, 1995), and the perception-production co-evolve pattern (Flege & Bohn, 2021). The perception and production of this target were completed with a ceiling accuracy score when the experiments were initiated and showed no correlation during the developmental procedure.
Regarding Research Question (4), perception training was empirically testified to be effective in promoting Mandarin speakers' learning of targets. The perception and production of Cantonese syllable-final segments by Mandarin speakers performed a gradual improvement during the intervention (Experiment 3). This trained group then outperformed the untrained group in Experiment 4 (the post-test).
Collectively, the significant findings of the current study are as follows. Empirically, this dissertation provides developmental data about Mandarin speakers' perception and production learning of seven Cantonese syllable-final segments which are undergoing a coronal-dorsal merging tendency ([-t]-[-k] and [-n]-[-ŋ]). Theoretically, the notion of cross-L1-L2 similarity in SLM-r and its predictions deduced by the perceptual similarity are generally verified in the current study. Different tenets employed by PAM-L2 and L2LP provide alternative accounts of the observed results beyond the SLM-r predictions. Furthermore, the effects of the distribution patterns of L2 targets are heuristically proposed to supplement the decisive role of cross-L1-L2 similarity in predicting L2 speech learning by the aforementioned L2 learning models. As for the relationship between L2 perception and production, diverse patterns of perception-production interface varying according to the learnability of different targets is advanced to replace one-size-fits-all patterns proposed by previous studies.
Rights: All rights reserved
Access: open access

Files in This Item:
File Description SizeFormat 
6224.pdfFor All Users2.49 MBAdobe PDFView/Open

Copyright Undertaking

As a bona fide Library user, I declare that:

  1. I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
  2. I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
  3. I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.

By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.

Show full item record

Please use this identifier to cite or link to this item: