How does contact with English Language influence the pronunciation of final consonants (-k, -t, -p) in Cantonese among bilingual speakers in ethnic enclaves/diaspora communities?
Introduction
Cantonese language, a variety of Yue Chinese dialects, is a Chinese language that belongs to the Sino-Tibetan language family. It originated in the city of Canton (Guangzhou) where a vital role in politics, economy, and culture were played throughout ancient times in the Southern Chinese province of Guangdong and maintained an official status in Hong Kong and Macau to this day (Egerod, 2025). Due to historical migration, Cantonese is spoken overseas by ethnic enclaves which are “neighborhoods that have a population that is largely ethnically distinguished from the surrounding area and its inhabitants” (Li, 2005) such as “Chinatown” within Vietnam, Malaysia, Singapore in Southeast Asia; United States and Canada in North America; and the United Kingdom and France in Europe. According to Ethnologue, there are approximately 85 to 90 million Cantonese speakers worldwide (Eberhard, Simons, & Fenning, 2019). According to the 2016 census of the Australian Bureau of Statistics, Cantonese was the fourth most spoken language by a person at home after English, Mandarin, and Arabic (Australian Bureau of Statistics, 2017). In addition, The Office of the Commissioner of Official Languages in Canada stated that Cantonese was the fourth most spoken language by a person at home after English, French, and Mandarin (Office of the Commissioner of Official Languages, 2018). As for the United States, the United States Census Bureau released data in 2015 that shows nearly 80% of the population only speak English with Chinese, including Mandarin and Cantonese, ranking the third most spoken language at home (United States Census Bureau, 2015). The presence of a large Cantonese-speaking population indicates significant Chinese immigration to these countries with Cantonese being the primary language of the immigrant population. It continues to serve as a primary community and heritage language for many of these established diaspora communities.
In the United States, the coexistence of Cantonese and English languages in ethnic enclaves creates sustained opportunities for language contact where linguistic characteristics of each language influence the other. Retaining majority features of Ancient Chinese, Cantonese has most of the final consonant pronunciation of Ancient Chinese and a complete set of nine tones compared to the four tones of Modern Standard Chinese. Cantonese has six final consonants: [p], [k], [t], [m], [n], and [ŋ] (The Education University of Hong Kong). This study will primarily focus on the final consonant pronunciation of [p], [k], and [t] as its final plosives are not audibly released. In contrast, English has abundant final consonants, categorized by velars, fricatives, and voiceless stops (The University of Kansas). The pronunciation of final consonants in English grammar is determined by the voicing of the preceding sound. A voiceless consonant is a consonant sound made without the vocal cord vibrating such as [p], [t], and [k]. To produce a voiceless consonant, air passes through the mouth without the vocal folds closing and buzzing, pronouncing with forceful puff of air. For Cantonese English as a Second Language (ESL) learners, they tend to apply the phonological rules from Cantonese to English pronunciation, a lack of or, perhaps, overemphasis on the final consonant pronunciation. In this context, confusion and communication barriers may arise in conversations.
In diaspora communities, contact with English is likely to cause Cantonese bilingual speakers to exhibit variation or reduction in the pronunciation of final stop consonants (-k, -p, -t), potentially leading to increased elision (deletion) or substitution with glottal stops. Speakers may converge their Cantonese pronunciation towards English norms, resulting in reduction or loss of the unreleased quality of final stops. In addition, the pressure from the English sound system may increase the rate of final consonant deletion in Cantonese, a feature sometimes associated with less formal or ethnolectal speech patterns. This variation is highly influenced by social factors, identity, and the relative prestige of English in their environment. Speakers may consciously or unconsciously use these variants to index their social identity, such as signaling a stronger orientation towards their ethnic in-group by maintaining “authentic” Cantonese features or toward a local, mult-ethnic identity by adopting a local ethnolect influenced by English. In studying this linguistic phenomenon, it provided a concrete case study of how language contact drives sound change and how features from a dominant language (English) can influence a heritage language (Cantonese). The research explored the complex interaction between ethnic and regional identities and language use. Pronunciation choices often reflect speakers' attitudes, loyalty to their heritage, and their integration into the wider society. This research helped fill gaps in the understanding of how heritage bilinguals maintain or shift aspects of their sound systems across generations in diaspora communities. It contributed to the study of how unique ethnic dialects (ethnolects) form, combining features of both the heritage language and the surrounding English variety. Understanding these specific sound variations can inform speech-language pathology practices and language education, helping professionals distinguish between a natural contact-induced variation and a language disorder. The following research question is proposed: How does contact with English Language influence the pronunciation of final consonants (-k, -t, -p) in Cantonese among bilingual speakers in ethnic enclaves/diaspora communities?
Literature Review
Historical Significance of the Cantonese Language
Amidst the history of Cantonese, political and social aspects have significantly influenced the transmission and evolution of the language. After the British acquired Hong Kong in 1842, throughout most of the colonial era, English was the sole official language of administration, law, and higher degree (Clark, 2018). This has established a clear diglossia: English was the high-prestige language, and Cantonese was the everyday vernacular. Long-term exposure to English has directed Cantonese linguistic evolution through extensive English borrowings, widespread code-switching (the practice of alternating between two or more languages), and phonological differences. In a research on Cantonese-English code-switching in Hong Kong, Dr. David Li, a professor and head of the Department of Chinese and Bilingual Studies at the Hong Kong Polytechnic University, predominantly identified the phenomenon as intrasentential code-mixing within a single sentence (Li, 2000). Social motivations for code-switching included communicative needs and using English for euphemism, specificity, bilingual punning, and economy, as well as displaying a “Western outlook” or a distinct “Hong Konger” identity. Following the 1997 handover of Hong Kong to China led to a shift in language policy in making Mandarin an official language alongside Cantonese and English in the Biliteracy and Trilingualism Policy. In a publication by the Chinese University of Hong Kong (CUNK), Yeung examines the sociolinguistic shift within the language dynamics in Hong Kong and the resistance against perceived mainlandisation. The paper concluded that there is an increased incorporation of Mandarin vocabulary into Cantonese, viewed by radicals as a cultural invasion that threatens the distinctiveness of Hong Kong identity (Yeung, 2021). As time progressed to 2014, the Hong Kong Umbrella Movement emerged as a protest for greater political freedom and universal suffrage. According to Shum et al. from the Hong Kong Polytechnic University, the 2014 Umbrella Movement led to Hong Kongers developing a strong preference for Cantonese and increased resistance to Mandarin, viewing Cantonese as in-group marker of Hong Kong identity and Mandarin as out-group language associated with political motivation for linguistic assimilation (Shum et al., 2023).
Ethnic Enclave
- 1. Historical Development
Steven J. Gold, a sociology professor at the Michigan State University and a sociologist researched in ethnic communities, ethnic economies, and international immigration, defines the term “Ethnic Enclave” as limited geographically regions where an immigrant community shares common national or ethnic origins (Gold, 2015). Drawn by the 1848 Gold Rush and railroad construction, the Chinese people were prompted to immigrate to California for economic opportunities. In the 19th century, Chinatown in the United States developed as Chinese immigrants faced hostility, economic hardship, and the 1882 Chinese Exclusion Act, a law that barred Chinese laborers from entering the U.S. and prevented existing laborers from obtaining citizenship (Library of Congress, n.d.). Following the completion of railroad construction and decline of the Gold Rush, many Chinese immigrants moved into self-reliant, urban neighborhoods in which they could shop for familiar groceries, worship traditional temples, and maintain cultural ties to their homelands. In addition, Chinese-owned businesses offered economic opportunities to Chinese workers where mainstream employers of non-Chinese businesses would not at the time. From there, Chinatown had expanded and spanned from the Western coast cities to the Eastern coast cities in forming a continental network.
- 2. Coexistence of languages within Chinatowns
The ethnic enclaves of the Chinese immigrants function as multilingual spaces where heritage language is both preserved and reshaped under majority-language pressures. In a qualitative study published in New Perspectives on Languages, a peer-reviewed journal, Tian & Kroon, analyzed more than 1,200 signs in Los Angeles Chinatown, found that 41 percent were monolingual Chinese, 32 percent bilingual Chinese-English, and 27 percent monolingual English. Moreover, there were evident competitions in Chinese forms: (1) orthographic choice: traditional and simplified Chinese are both present, (2) romanization systems: standardization and local tradition existed between Hanyu pinyin and older romanization practices, and (3) vocabulary choice: Putonghua, Cantonese, and Taiwan Mandarin coexist which reflects multilingual and multicultural nature of diaspora community (Tian & Kroon, 2024). Adding on with a case study in London Chinatown, Harland, Pan, & Barnes documented that 68 percent of signage deliberately retained Chinese characters alongside English, suggesting a practice named “graphic heritage” to project cultural identity and continuity (Harland, Pan, & Barnes, 2024). While competition and adaptation were showcased in the study of the Los Angeles Chinatown, the case in London Chinatown emphasized heritage preservation. Ethnic enclaves are dynamic sites where language contacts are interchangeable in balancing accessibility and commercial appeal, selected for heritage visibility in multicultural urban spaces.
Language Contact and Adoption
In Hong Kong’s unique sociolinguistic landscape, the interaction between Cantonese, English, and Mandarin is characterized by complex language contact and adaptation where different generations of speakers navigate through dynamic environments of change and maintenance. According to the study published by Dr. Picus S. Ding, professor and university scholar at the School of Humanities of the University of Hong Kong, concluded that Hong Kong Cantonese (HKC) has been influenced by both Chinese topolects and English, resulting in specific changes like the assimilation of certain sounds, the adoption of new syllable structures, and the merger of new tones (ex. [-n] into [-l]). English has introduced new syllable types, including consonant clusters, and contacts with other Chinese dialects has introduced phonetic variations (Ding, 2010). The subsequent journal article published in Journal of Association for Laboratory Phonology operationalizes the observations of Ding (2010) using modern phonetic analysis. The 2022 paper by Cheng, Babel, and Yao provides empirical evidence for the status of several mergers Ding mentioned in that they are active changes in the community using quantitative data on production and perception across different age groups. The study found that the [n] to [l] merger in Hong Kong Cantonese is nearly complete, with over 90% production of [l] across age groups, though younger listeners are more perceptually sensitive to the distinction. The [ŋ] to [m] merger is also well underway, while the [ŋ] to Ø merger shows less robust group-level contrast, with a mean production of the null-initial form at 24% across groups (Cheng et al., 2022). On another perspective, Soo, Babel, and Johnson, researchers from the University of British Columbia, Canada, analyzes the fine-grained cross-linguistic phonetic variation within bilingual individuals concerning the [n] to [l] merger. The study investigates subtle acoustic differences in how early Cantonese-English bilingual speakers produce the English [n] to [l] contrast versus the variable Cantonese contrast (Soo et al., 2024). This paper is oriented on whether the phonetic characteristics of the sounds speakers make in English influence their production of the sounds in Cantonese.
Gap Statement
Although a substantial body of literature explores linguistic changes in Cantonese, much of the existing research has centered on tone mergers, code-switching, or lexical borrowing in general urban dialect shifts in Hong Kong. Studies on language contact and bilingual interference often focus on the influence of Mandarin or other Chinese dialects which neglects the specific features of English as a contact language in Western diasporas. Furthermore, heritage language maintenance in the United States Asian communities often remains at a macro-soclingistic level in examining language choice, domain use, and evaluating its overall vitality. Rarely, the maintenance work would analyze small speech details of subphonemic shifts occurring in production. This overlooking omission results in a significant research gap that there is distinct lack of research directly exploring how to absence of final stops in English language influences the production and maintenance of of Cantonese consonants (-k, -t, and -p) among bilingual Chinese-American speakers living in diaspora communities. This study is specifically designed to address this gap to contribute on the topic of diaspora sociophonetics by providing in-depth, production-based phonological analysis of the bilingual speaker population. Therefore, the study provided a comprehensive understanding of mechanisms of sound change under intense language contact.
Rationale
To examine the contact between English and Cantonese in influencing the pronunciation of final consonants in Cantonese, quantitative acoustic analysis is the most appropriate method for this study. By inputting audio samples in Cantonese from both bilinguals and monolinguals into Praat, a free and accessible computer software program widely used for speech analysis in phonetics, I made precise acoustic visualization of closure duration, release bursts, and burst intensity in which are often unreleased or subtle through inspecting the waveform and spectrogram of segmented speech (see Appendix B and C for all spectrograms). Descriptions of each variable are shown in Table 1.
Table 1. Definitions of the analyzed variables.
| Variable | Definition |
| Release vs. unreleased final consonants | A stop is released when the oral closure is followed by an audible burst of air; it is unreleased when the closure is maintained without a burst and the stop ends in silence (Ladefoged & Johnson, 2014). |
| Closure Duration | The interval during which the vocal tract is fully occluded for a stop consonant, measured from the end of the preceding vowel to the onset of the burst or release (Johnson, 2012). |
| Burst Duration | The duration of the transient noise produced when the stop closure is released, visible as a brief high-energy spike in the waveform and spectrogram (Stevens, 1998). |
| Burst Intensity | The amplitude of the stop burst, measured in decibels (dB). This variable shows the pressure buildup and force of the release (Kent & Read, 2002). |
Out of all of the Cantonese corpus, I chose to focus on the SpiCE corpus because it provides high-quality bilingual data and Cantonese has been used for a long time in Vancouver where the corpus was designed. More specifically, it consists of conversational Cantonese-English speech recorded in the Cantonese-speaking community of Metro Vancouver between 2018 and 2020 of 34 early bilinguals participating in tasks, including interviews, sentence reading, and storyboard. Since the large corpus contains hand-corrected orthographic transcriptions and phoneme-level forced alignment, this facilitated in locating the targeted final consonants of Cantonese words (Table 2) efficiently. The MagicHub Cantonese corpus provides a monolingual comparison group. It contained conversational speech from native Cantonese speakers and recorded in naturalistic settings. The dataset included demographic metadata and phonetic transcriptions, and it is designed to represent everyday Cantonese usage. This corpus is also large with thousands of utterances and is suitable for extracting tokens per speaker. Considering that the recordings were collected under controlled conditions in terms of spontaneous speech, metadata, and high-quality audio for acoustic segmentation, the analysis is replicable and transparent.
Research Design
The study adopts a comparative design approach between two speaker groups: Cantonese-English bilinguals and Cantonese monolinguals. For each group, 10 speakers are included, totaling 20 speakers. Bilingual speakers are selected from the SpiCE corpus, and monolingual speakers from the MagicHub Corpus. Inclusion criteria is based on the availability of sufficient tokens of final /ptk/ in spontaneous speech. For each speaker, six tokens of word-final stop are analyzed: two tokens for each of the three target consonants, -p, -t, and -k. When a specific Cantonese word containing a target final consonant is not available for a given speaker, an alternative containing the same final consonant in coda position is selected instead.
Table 2. A list of targeted Cantonese words, Jyutping (standardized romanization system), and its meaning in English.
| Final Consonant | Word | Jyutping | Meaning in English |
| 福 | fuk1 | Blessing; good fortune; luck | |
| 學 | hok6 | Learning; knowledge | |
| K | 六 | luk6 | Six |
| 國 | gwok3 | Country | |
| 食 | sik6 | Eat; have a meal | |
| 一 | jat1 | One | |
| T | 失 | sat1 | Mistake; lose; neglect |
| 發 | faat3 | To send out, develop, or make money | |
| 日 | jat6 | Day; Sun | |
| 入 | jap6 | Enter; come/go in(to) | |
| 十 | sap6 | Ten | |
| P | 合 | hap6 | Close; join; whole |
| 盒 | hap6 | Small box; case | |
| 業 | jip6 | Profession; business; industry; studies | |
| 習 | zaap6 | Learn; receive training in; practice |
Data Collection
The extraction of acoustic measurements in Praat are carried out in a consistent, step-based manner that is applied to all tokens across both groups. In analyzing closely into the releases, closure duration, bursts, and burst intensity, it is imperative to follow a standardized workflow to ensure comparability across speakers and corpora. The following flowchart (Figure 1) summarized the sequence of procedures performed for each token.
Figure 1. Flowchart of procedures conducted in Praat.
Data Analysis
The acoustic measurements extracted from Praat were completed into a spreadsheet for statistical analysis (see Appendix A for the full spreadsheet). Each speaker contributed six tokens, two tokens of each final consonant, yielding 60 tokens per group and in total 120 tokens. For each token, four variables were recorded: released/unreleased (1/1 or 0/1), closure duration (seconds), burst duration (released tokens only), and burst intensity (released tokens only). All analyses are performed in Microsoft Excel. The following measures will be computed for closure duration, burst duration, and burst intensity: Mean, Standard Deviation, and number of tokens contributing to each measure. Release frequency is summarized using a count-based table. For each group, the number of released tokens are tallied separately for /p/, /t/, and /k/, and a total release count is calculated. This table provided a categorical comparison of release patterns across groups and across consonant types. Using Excel’s built-in programming, group differences are evaluated using independent-samples t-tests for each variable. Statistical significance was evaluated at p<0.05 level of significance. To visualize the properties of the acoustic variables across the two groups, three box and whiskers plots were created: Closure Duration: Bilingual vs. Monolingual, Burst Duration: Bilingual vs. Monolingual, and Burst Intensity: Bilingual vs. Monolingual. Each plot displayed the median, interquartile range, and outliers for the two groups.
Results
Table 1: Count of Release
Table 1 below presents the distribution of released (1) and unreleased (0) final stops across bilingual and monolingual speakers, with counts further broken down by place of articulation (K, P, T). A total of 120 tokens were analyzed, with 60 tokens produced by each group. Across all speakers, 55 tokens were unreleased and 65 were released. The bilingual group produced 39 released tokens, while the monolingual group produced 26. Released counts varied by place of articulation within each group. The 10 bilingual speakers produced 11 released K tokens (55%), 10 released P tokens (50%), and 18 released T tokens (90%), for a total of 39 released tokens (65%) across all places. Within the monolingual group, nine released K tokens (45%), eight released P tokens (40%), and nine released T tokens (45%) were observed, totaling 26 released tokens (43.3%).
Figure 1: Closure Duration
Figure 1 displayed the distribution of closure duration values for bilingual and monolingual speakers. Closure duration values ranged from approximately 0.03 to 0.45 seconds across all tokens. The boxplots showed comparable median closure durations for the two groups, both slightly below 0.1 seconds. The bilingual group exhibited a narrower interquartile range, while the monolingual group showed a wider spread of values. Each group contained one high-duration outlier above 0.25 seconds, with the monolingual outlier exceeded 0.40 seconds. An independent sample t-test was performed to compare the average closure duration of the bilingual group (M = 0.0695 s (69.5 ms), SD = 0.0399) versus the monolingual group (M = 0.0901 s (90.1 ms), SD = 0.0535). There was a significant difference in the closure duration values; p = 0.018.
Figure 2: Burst Duration
Figure 2 presented mean burst duration values for released tokens of bilingual and monolingual speakers, accompanied by individual data points and error bars. The bilingual group showed burst durations distributed around a lower mean value, with individual points forming a relatively tight cluster. Error bars represented a modest degree of within-group variability. The monolingual group displayed a higher mean burst duration, with individual data points spread across a wider vertical range. Error bars showed greater variability compared to the bilingual group. The distribution of points around each bar illustrated the dispersion of burst duration values within each group. An independent sample t-test was performed to compare the average burst duration of the bilingual group (M = 0.0319 s (31.9 ms), SD =0.0200 ) versus the monolingual group (M = 0.0481 s (48.1 ms), SD = 0.0199). There was a significant difference in the burst duration values; p = 0.002.
Figure 3: Burst Intensity
Figure 3 illustrated the mean burst intensity values in decibels of released tokens for bilingual and monolingual speakers. The bilingual group showed intensity values distributed around a lower mean, with individual data points forming a moderately dispersed cluster, Error bars indicated the range of variability within the group. The monolingual group showed a higher mean burst intensity, with individual data points distributed across a similar vertical range. Error bars reflected within-group variability comparable to that of the bilingual group. The plotted points showed the spread of intensity values around each group’s mean. An independent sample t-test was performed to compare the average burst intensity of the bilingual group (M = 60.79 dB, SD = 5.52) versus the monolingual group (M = 63.41 dB, SD = 5.91). There was an insignificant difference in the burst intensity values; p = 0.078.
Discussion
Based on the results gathered, the patterns observed contribute to a wider understanding of bilingual phonetic behavior. The statistical outcomes provide a foundation for interpreting how Cantonese-English bilinguals differ from monolingual Cantonese speakers in their production of final stops. Bilingual speakers produced noticeably more released final consonants than monolingual speakers, especially for /t/, where the bilingual release rate reached 90%. A potential explanation for this pattern is the influence of English phonetic contact, which favors audible releases in final stops. This trend aligns with previous research by Kong (2009) and Spinu et al. (2023), who found that bilinguals often release final stops more frequently than monolinguals. It also fits within broader discussions of L2 to L1 phonetic transfer (Chang, 2012). This suggests that speakers with bilingual experience can subtly alter native-language articulation. On the other hand, the bilingual group produced significantly shorter closure durations than the monolingual group. This outcome is mostly resulted from cross-linguistic interactions. By Bauer & Benedict (1997), English final stops typically occur with shorter closures, while Cantonese final stops are known for longer, unreleased closures. The shorter closures of the bilinguals therefore suggest some degree of English interference on the Cantonese closure patterns. Although, both groups showed a few unusually long closures (outliers) which suggests individual variation or momentary articulation. A larger dataset would help clarify whether this is a stable bilingual effect or a more rigid trend. Given a strong statistical difference, bilingual speakers also produced shorter burst durations than monolingual speakers. This trend may be attributed to the fact that bilinguals released more stops overall. More frequent releases can lead to quicker, lighter bursts on the basis that speakers are transitioning between speaking habits of Cantonese and English. This pattern resembles findings by Spinu et al. (2023) in which they concluded that bilingual speakers often show reduced burst durations due to faster articulatory transitions. In addition, the shorter burst durations can be explained by the Speech Accommodation Theory – a notion where individuals may adjust their communication styles to align phonetic patterns across languages in order to minimize social differences with others. Regarding burst intensity, monolingual speakers showed slightly higher burst intensity values than bilinguals, but the difference was not statistically significant. While not significant, the small increase in monolingual intensity could imply that when monolingual speakers do release a final consonant, the release may be forceful. This contrasts with the study of Yiu (2019) who found stronger bursts among bilinguals, but differences in task design or token selection could make a difference in the data. It is also possible that the lack of significance reflects the smaller number of released tokens in the monolingual group. However, these conclusions must be interpreted with caution due to intensity being highly sensitive to recording conditions.
Taken together, the results show that bilingual speakers produce Cantonese final stops in ways that differ from the monolinguals, aligning to the English phonetic patterns. This contributes to the growing understanding that bilingual phonetic systems are dynamic rather than fixed. On a broader level, these results raise questions about how stable phonetic categories are in adulthood. Several limitations remain for the sample size was modest and only three places of articulation were examined. Future research should incorporate more spontaneous speech, additional acoustic measures, and sociolinguistic variables such as language dominance and age range of the participants.
Conclusion
This study examined how Cantonese-English speakers differ from Cantonese monolinguals in their production of final stop consonants and focus on release frequency, closure duration, burst duration, and burst intensity. By conducting acoustic measurements in Praat using two established corporas, the research aimed to provide a clearer understanding of how bilingual experience shapes the phonetic realization of segments in a language where unreleased codas are the traditional convention. Across all measures, the most striking difference between the two groups was the frequency of released final stops. Bilinguals released a noticeably higher production of final stops than monolinguals which means that English may influence how bilinguals articulate Cantonese stops. This finding suggested broader claims in bilingual phonetics that speakers often carry over articulatory habits from one language to another. Bilingual speakers produced significantly shorter closure durations and shorter burst durations than monolingual speakers. These differences point toward a shift in articulatory timing that aligns more closely with English stop production. The consistency of these findings strengthens the interpretation that bilinguals may be adopting intermediate timing patterns influenced by both languages. These results contribute to ongoing discussion in literature about how bilinguals manage different phonetic systems and how such management offers acoustic outcomes. Burst intensity, however, did not show a statistically significant difference between groups. Although monolingual speakers displayed a slightly higher mean intensity, the overlap between the two groups suggests that intensity may be less sensitive to cross-linguistic influence than timing-based measures. Not all acoustic measures shift in the same direction, and some may remain relatively stable despite exposure to another language.
2006 2017, Census All persons QuickStats | Australian Bureau of Statistics. (2017). Abs.gov.au; Australian Bureau of Statistics. https://abs.gov.au/census/find-census-data/quickstats/2006/POA2017
Bauer, R. S., & Benedict, P. K. (1997). Modern Cantonese Phonology. Walter de Gruyter.
Bureau, U. C. (2022, July). What Languages Does the United States Speak? A Geographic Analysis of the Languages Spoken at Home in the United States, 2015-2019. Census.gov. https://www.census.gov/library/visualizations/2022/demo/lang-geogr-distrob.html
Castree, N., Rogers, A., & Kitchin, R. (2013). A Dictionary of Human Geography. Oxford University Press. (Original work published 2013)
Cheng, L. S. P., Babel, M., & Yao, Y. (2022). Production and perception across three Hong Kong Cantonese consonant mergers: Community- and individual-level perspectives. Volume 13, 13(1). https://doi.org/10.16995/labphon.6461
Ding, P. (2010). Phonological change in Hong Kong Cantonese through language contact with Chinese topolects and English over the past century.
Eberhard, D. M., Simons, G. F., Fennig, C. D., & Institute, S. (2019). Ethnologue. Languages of Asia. Sil International.
Gold, S. J. (2015). Ethnic Enclaves. Emerging Trends in the Social and Behavioral Sciences, 1–18. https://doi.org/10.1002/9781118900772.etrds0118
Government of Canada. (2024, August 14). Statistics on official languages in Canada. Government of Canada. https://www.canada.ca/en/canadian-heritage/services/official-languages-bilingualism/publications/statistics.html
Harland, R. G. (2024). The heritage language and graphic landscape of London’s Chinatown. Proceedings of DRS. https://doi.org/10.21606/drs.2024.885
Hu, J. (2022). Metrolingual Linguistic Repertoire and Business Sovereignty: Geosemiotic Analysis of Linguistic Landscapes in Sydney Chinatown. Advances in Social Science, Education and Humanities Research. https://doi.org/10.2991/assehr.k.220401.034
Li, D. C.-S. (2000, November). (PDF) Cantonese-English code-switching research in Hong Kong: A Y2K review. ResearchGate. https://www.researchgate.net/publication/227627801_Cantonese-English_code-switching_research_in_Hong_Kong_A_Y2K_review
Li, W. (2005). Beyond Chinatown, beyond Enclave: Reconceptualizing Contemporary Chinese Settlements in the United States. GeoJournal, 64(1), 31–40. https://doi.org/10.1007/s10708-005-3921-6
Library Of Congress. (n.d.). Building Communities | Chinese | Immigration and Relocation in U.S. History | Classroom Materials at the Library of Congress | Library of Congress. Library of Congress, Washington, D.C. 20540 USA. https://www.loc.gov/classroom-materials/immigration/chinese/building-communities/
Magic Data. (2019). Magicdatatech.com. https://www.magicdatatech.com/datasets/asr/mdt-asr-e018-hong-kong-cantonese-speech-corpus-1608279284
Pronunciation of Hong Kong English – Phonetics and Phonology. (n.d.). The Education University of Hong Kong. https://corpus.eduhk.hk/english_pronunciation/index.php/pronunciation-of-hong-kong-english/
Shum, P. L.-C., & Hamamura, T. (2023, June). The Effects of Large-Scale Social Movements on Language Attitudes: Cantonese and Mandarin in Hong Kong (C.-S. Tse & S. Wright, Eds.). ResearchGate. https://www.researchgate.net/publication/367225488_The_Effects_of_Large-Scale_Social_Movements_on_Language_Attitudes_Cantonese_and_Mandarin_in_Hong_Kong
Soo, R., Babel, M., & Johnson, K. A. (2024). Cross-Linguistic Phonetic Variation in Bilingual Speech: Cantonese /n/ > [l] Merger in Early Cantonese–English Bilinguals. Language and Speech. https://doi.org/10.1177/00238309241280182
SpiCE Corpus. (2021). Readthedocs.io. https://spice-corpus.readthedocs.io/en/latest/#citing-the-corpus
Spinu, L., Hwang, J., & Vasilita, M. (2023). Differences between Monolinguals and Bilinguals in Phonetic and Phonological Learning and the Connection with Auditory Sensory Memory. Brain Sciences, 13(3), 488. https://doi.org/10.3390/brainsci13030488
States, U. (2026). Explore Census Data. Census.gov. https://data.census.gov/table/ACSDT5Y2015.B16001?t=Language+Spoken+at+Home&g=040XX00US31
Théberge, R. (2018). Office of the Commissioner of Official Languages (Canada). Languagecommissioners.org; International Association of Language Commissioners. https://languagecommissioners.org/Office-of-the-Commissioner-of-Official-Languages
The University of Kansas. (n.d.). Chapter 2 – Consonants. Opentext.ku.edu. https://opentext.ku.edu/amenglishpronunciation/chapter/consonants-and-word-level-grammar/
Tian, F., & Kroon, S. (2025). View of The Linguistic Landscape of Los Angeles Chinatowns. At-Journals.com. https://at-journals.com/journals/index.php/IJSEFOL/article/view/10/12
Yeung, K. C. L. (2021). On the Mainlandisation of Cantonese: Language and Identity. Hong Kong Anthropologist, 11(2). https://www.arts.cuhk.edu.hk/~ant/hka/documents/2021/HKA11_LennoxYEUNG.pdf
Yiu, Y. C. (2019). Directional Particles in Cantonese: Form, Function, and Grammaticalization. Studies in Chinese Language Discourse 9. Jstor.org, 47(1). https://doi.org/10.2307/26705058
Zeng, J., & Agha, A. (2021). Cantonese as a World Language. Journal of Student Research, 10(2). https://doi.org/10.47611/jsrhs.v10i2.1435
Appendix A
- 1. Data recorded on Excel for variables (release, closure duration, burst duration, and burst intensity) collected from acoustic analysis on Praat
| Speaker ID | Group (Bi/Mono) | English Use Level | Word | Final Stop | Release (0/1) | Closure Duration (s) | Burst Duration (s) | Burst Intensity (dB) |
| BI01 | Bi | 福 (fuk 1) | K | 0 | 0.125 | NA | NA | |
| BI01 | Bi | 學 (hok 6) | K | 1 | 0.073 | 0.058 | 63.56 | |
| BI01 | Bi | 發 (faat 3) | T | 1 | 0.05 | 0.029 | 62.59 | |
| BI01 | Bi | 一 (jat 1) | T | 1 | 0.04 | 0.023 | 67.53 | |
| BI01 | Bi | 業(jip 6) | P | 1 | 0.15 | 0.076 | 61.15 | |
| BI01 | Bi | 十 (sap 6) | P | 1 | 0.09 | 0.08 | 52.98 | |
| BI02 | Bi | 福 （fuk 1) | K | 0 | 0.04 | NA | NA | |
| BI02 | Bi | 食 （sik 6) | K | 0 | 0.12 | NA | NA | |
| BI02 | Bi | 發 （faat 3) | T | 1 | 0.012 | 0.0106 | 65.06 | |
| BI02 | Bi | 一 （jat 1 ) | T | 1 | 0.07 | 0.07 | 64.04 | |
| BI02 | Bi | 業 （jip 6) | P | 0 | 0.13 | NA | NA | |
| BI02 | Bi | 盒 （hap 6) | P | 1 | 0.11 | 0.05 | 50.17 | |
| BI03 | Bi | 福 （fuk 1) | K | 0 | 0.14 | NA | NA | |
| BI03 | Bi | 學 （hok 6 ) | K | 1 | 0.07 | 0.03 | 55.37 | |
| BI03 | Bi | 發 （faat 3) | T | 1 | 0.04 | 0.02 | 61.71 | |
| BI03 | Bi | 一 （jat 1 ) | T | 1 | 0.03 | 0.02 | 68.15 | |
| BI03 | Bi | 業 （jip 6) | P | 1 | 0.07 | 0.028 | 57.9 | |
| BI03 | Bi | 習 (zaap 6) | P | 1 | 0.09 | 0.04 | 56.17 | |
| BI04 | Bi | 福 （fuk 1) | K | 0 | 0.06 | NA | NA | |
| BI04 | Bi | 食 （sik 6) | K | 1 | 0.03 | 0.02 | 57.07 | |
| BI04 | Bi | 發 （faat 3) | T | 1 | 0.07 | 0.048 | 62.43 | |
| BI04 | Bi | 一 （jat 1 ) | T | 1 | 0.04 | 0.038 | 74.33 | |
| BI04 | Bi | 業 （jip 6) | P | 0 | 0.13 | NA | NA | |
| BI04 | Bi | 入 （jap 6) | P | 0 | 0.05 | NA | NA | |
| BI05 | Bi | 福 （fuk 1) | K | 0 | 0.11 | NA | NA | |
| BI05 | Bi | 食 （sik 6) | K | 0 | 0.03 | NA | NA | |
| BI05 | Bi | 發 （faat 3) | T | 1 | 0.11 | 0.028 | 60 | |
| BI05 | Bi | 一 （jat 1 ) | T | 1 | 0.05 | 0.046 | 73.42 | |
| BI05 | Bi | 業 （jip 6) | P | 0 | 0.23 | NA | NA | |
| BI05 | Bi | 入 （jap 6) | P | 0 | 0.13 | NA | NA | |
| BI06 | Bi | 福 （fuk 1) | K | 1 | 0.1 | 0.011 | 53.64 | |
| BI06 | Bi | 六 （luk 6) | K | 1 | 0.06 | 0.017 | 57.5 | |
| BI06 | Bi | 發 （faat 3) | T | 1 | 0.08 | 0.025 | 57.12 | |
| BI06 | Bi | 一 （jat 1 ) | T | 1 | 0.04 | 0.013 | 65.34 | |
| BI06 | Bi | 業 （jip 6) | P | 0 | 0.011 | NA | NA | |
| BI06 | Bi | 入 （jap 6) | P | 0 | 0.06 | NA | NA | |
| BI07 | Bi | 福 （fuk 1) | K | 0 | 0.04 | NA | NA | |
| BI07 | Bi | 食 （sik 6) | K | 0 | 0.04 | NA | NA | |
| BI07 | Bi | 發 （faat 3) | T | 1 | 0.04 | 0.022 | 57.83 | |
| BI07 | Bi | 一 （jat 1 ) | T | 1 | 0.03 | 0.022 | 58.41 | |
| BI07 | Bi | 入 （jap 6) | P | 1 | 0.08 | 0.013 | 53.29 | |
| BI07 | Bi | 業 （jip 6) | P | 1 | 0.07 | 0.018 | 50.78 | |
| BI08 | Bi | 福 （fuk 1) | K | 1 | 0.09 | 0.088 | 53.8 | |
| BI08 | Bi | 學 （hok 6 ) | K | 1 | 0.04 | 0.025 | 60.56 | |
| BI08 | Bi | 一 （jat 1 ) | T | 1 | 0.03 | 0.028 | 64.9 | |
| BI08 | Bi | 日 （jat 6) | T | 0 | 0.06 | NA | NA | |
| BI08 | Bi | 業 （jip 6) | P | 1 | 0.07 | 0.033 | 59.36 | |
| BI08 | Bi | 合 （hap 6) | P | 1 | 0.06 | 0.017 | 61.25 | |
| BI09 | Bi | 福 （fuk 1) | K | 1 | 0.1 | 0.02 | 60.09 | |
| BI09 | Bi | 學 （hok 6 ) | K | 1 | 0.03 | 0.027 | 66.49 | |
| BI09 | Bi | 發 （faat 3) | T | 1 | 0.03 | 0.017 | 63.39 | |
| BI09 | Bi | 一 （jat 1 ) | T | 1 | 0.07 | 0.023 | 63.78 | |
| BI09 | Bi | 業 （jip 6) | P | 0 | 0.07 | NA | NA | |
| BI09 | Bi | 入 （jap 6) | P | 0 | 0.05 | NA | NA | |
| BI10 | Bi | 學 （hok 6 ) | K | 1 | 0.06 | 0.057 | 64.02 | |
| BI10 | Bi | 食 （sik 6) | K | 1 | 0.1 | 0.012 | 59.53 | |
| BI10 | Bi | 發 （faat 3) | T | 0 | 0.06 | NA | NA | |
| BI10 | Bi | 一 （jat 1 ) | T | 1 | 0.04 | 0.027 | 60.19 | |
| BI10 | Bi | 入 （jap 6) | P | 0 | 0.04 | NA | NA | |
| BI10 | Bi | 十 （sap 6) | P | 1 | 0.03 | 0.018 | 66.2 | |
| MO01 | Mo | 食 （sik 6) | K | 0 | 0.074 | NA | NA | |
| MO01 | Mo | 學 （hok 6 ) | K | 0 | 0.075 | NA | NA | |
| MO01 | Mo | 一 （jat 1 ) | T | 1 | 0.084 | 0.084 | 63.95 | |
| MO01 | Mo | 日 （jat 6) | T | 1 | 0.15 | 0.03 | 66.97 | |
| MO01 | Mo | 十 （sap 6) | P | 0 | 0.097 | NA | NA | |
| MO01 | Mo | 入 （jap 6) | P | 1 | 0.042 | 0.032 | 53.65 | |
| MO02 | Mo | 食 （sik 6) | K | 0 | 0.104 | NA | NA | |
| MO02 | Mo | 學 （hok 6 ) | K | 1 | 0.084 | 0.034 | 52.49 | |
| MO02 | Mo | 日 （jat 6) | T | 1 | 0.1 | 0.03 | 66.62 | |
| MO02 | Mo | 一 （jat 1 ) | T | 1 | 0.15 | 0.065 | 66.27 | |
| MO02 | Mo | 十 （sap 6) | P | 1 | 0.052 | 0.045 | 56.43 | |
| MO02 | Mo | 入 （jap 6) | P | 1 | 0.14 | 0.076 | 57.98 | |
| MO03 | Mo | 六 （luk) | K | 1 | 0.074 | 0.005 | 68.92 | |
| MO03 | Mo | 學 （hok 6 ) | K | 1 | 0.063 | 0.039 | 66.55 | |
| MO03 | Mo | 日 （jat 6) | T | 0 | 0.029 | NA | NA | |
| MO03 | Mo | 一 （jat 1 ) | T | 0 | 0.096 | NA | NA | |
| MO03 | Mo | 十 （sap 6) | P | 0 | 0.167 | NA | NA | |
| MO03 | Mo | 入 （jap 6) | P | 0 | 0.126 | NA | NA | |
| MO04 | Mo | 食 （sik 6) | K | 0 | 0.153 | NA | NA | |
| MO04 | Mo | 學 （hok 6 ) | K | 0 | 0.405 | NA | NA | |
| MO04 | Mo | 日 （jat 6) | T | 1 | 0.101 | 0.054 | 61.86 | |
| MO04 | Mo | 一 （jat 1 ) | T | 1 | 0.071 | 0.069 | 62.11 | |
| MO04 | Mo | 十 （sap 6) | P | 0 | 0.066 | NA | NA | |
| MO04 | Mo | 入 （jap 6) | P | 0 | 0.082 | NA | NA | |
| MO05 | Mo | 食 （sik 6) | K | 1 | 0.109 | 0.062 | 64.91 | |
| MO05 | Mo | 六 （luk) | K | 0 | 0.058 | NA | NA | |
| MO05 | Mo | 日 （jat 6) | T | 0 | 0.027 | NA | NA | |
| MO05 | Mo | 一 （jat 1 ) | T | 1 | 0.054 | 0.024 | 65.37 | |
| MO05 | Mo | 十 （sap 6) | P | 0 | 0.072 | NA | NA | |
| MO05 | Mo | 入 （jap 6) | P | 0 | 0.067 | NA | NA | |
| MO06 | Mo | 食 （sik 6) | K | 0 | 0.09 | NA | NA | |
| MO06 | Mo | 六 （luk) | K | 0 | 0.173 | NA | NA | |
| MO06 | Mo | 一 （jat 1 ) | T | 1 | 0.078 | 0.038 | 66.12 | |
| MO06 | Mo | 失 （sat 1) | T | 0 | 0.073 | NA | NA | |
| MO06 | Mo | 十 （sap 6) | P | 1 | 0.059 | 0.024 | 61.1 | |
| MO06 | Mo | 入 （jap 6) | P | 1 | 0.146 | 0.078 | 67.19 | |
| MO07 | Mo | 學 （hok 6) | K | 1 | 0.155 | 0.075 | 71.36 | |
| MO07 | Mo | 國 （gok) | K | 1 | 0.102 | 0.056 | 65.61 | |
| MO07 | Mo | 一 （jat 1 ) | T | 0 | 0.067 | NA | NA | |
| MO07 | Mo | 日 (yat) | T | 0 | 0.095 | NA | NA | |
| MO07 | Mo | 入 （jap 6) | P | 0 | 0.089 | NA | NA | |
| MO07 | Mo | 十 （sap 6) | P | 0 | 0.077 | NA | NA | |
| MO08 | Mo | 學 （hok 6) | K | 1 | 0.067 | 0.05 | 48.49 | |
| MO08 | Mo | 六（luk 6) | K | 0 | 0.095 | NA | NA | |
| MO08 | Mo | 一 （jat 1 ) | T | 0 | 0.077 | NA | NA | |
| MO08 | Mo | 日 (yat) | T | 0 | 0.041 | NA | NA | |
| MO08 | Mo | 入 （jap 6) | P | 1 | 0.047 | 0.023 | 67.67 | |
| MO08 | Mo | 十 （sap) | P | 0 | 0.063 | NA | NA | |
| MO09 | Mo | 學 （hok 6) | K | 0 | 0.048 | NA | NA | |
| MO09 | Mo | 六（luk 6) | K | 1 | 0.072 | 0.049 | 69.24 | |
| MO09 | Mo | 一 （jat 1 ) | T | 1 | 0.072 | 0.045 | 70.47 | |
| MO09 | Mo | 日 (yat) | T | 0 | 0.058 | NA | NA | |
| MO09 | Mo | 入 （jap 6) | P | 0 | 0.067 | NA | NA | |
| MO09 | Mo | 業 （jip 6) | P | 0 | 0.047 | NA | NA | |
| MO10 | Mo | 食 （sik 6) | K | 0 | 0.113 | NA | NA | |
| MO10 | Mo | 學 （hok 6) | K | 1 | 0.075 | 0.056 | 66.99 | |
| MO10 | Mo | 發( faat ) | T | 0 | 0.065 | NA | NA | |
| MO10 | Mo | 一 （jat 1 ) | T | 0 | 0.053 | NA | NA | |
| MO10 | Mo | 入 （jap 6) | P | 1 | 0.092 | 0.047 | 65.16 | |
| MO10 | Mo | 業 （jip 6) | P | 1 | 0.08 | 0.063 | 55.42 |
Appendix B
- 1. Screenshots of sample Praat spectrograms on token for bilingual speakers
|
 |
|
 |
|
 |
|
 |
|
 |
|
 |
|
 |
|
 |
Appendix C
- 1. Screenshots of sample Praat spectrograms on token for monolingual speakers
|
 |
|
 |
|
 |
|
 |
|
 |
|
 |
|
 |
