Presentation
Cross-linguistic comparisons of high vowels within individual speakers of Toronto Heritage Cantonese"
Workshop on Immigrant Languages in the Americas (WILA 13)
(2022)
Abstract
One motivation behind the Heritage Language Variation and Change (HLVC) in Toronto Corpus is to provide more rigorous support for contact-induced change by facilitating analyses involving multiple sets of comparisons in spontaneous speech samples (Nagy 2011). While many phonetic studies using the HLVC Corpus have included inter-generational and diatopic comparisons, cross-linguistic comparisons have largely been indirect or based on community-level patterns (cf. Hoffman & Walker 2010) rather than based on how two languages interact with each other within the same individual speakers.
The current study uses raw formant measurements from HLVC recordings to compare acoustic production of three high vowels: two in Cantonese (/y/ as in捲[kyn35] ‘roll’ and /u/ as in 館[kun35] ‘public building’) and one in English (/u/ as in ‘goose’ and henceforth ‘UW’). The questions addressed are:
Q1) Is UW (which has been described as phonetically fronted, cf. Boberg 2011) more similar to /y/ or /u/?
Q2) How does /y/~UW~/u/ production vary based on group (Homeland, Toronto Gen1, Toronto Gen2), Pillai Scores (to measure /y/~/u/ merger), and Cantonese Production Scores (a proficiency proxy)?
Participants were told that the interviews were to be primarily in Cantonese, but were allowed to code-switch into English as often as was natural. Only speakers producing at least 4 tokens of each of the three vowels were analyzed. This amounted to a total of 22 speakers (two Homeland, nine Gen1, and 11 Gen2) producing a total of 851 tokens of /y/, 525 tokens of /u/, and 270 tokens of UW. Regression models were run for each individual speaker with F2 as the dependent variable and vowel category (/y/ vs. UW or /u/ vs. UW) as the independent variable. Significant results were interpreted as phonetic distinction while non-significant results were interpreted as merger.
Overall results show individual speakers grouping into three distinct patterns:
P1) /y/, /u/, and UW are all distinct (n = 11)
P2) UW and /u/ are merged (n = 7)
P3) /y/ and UW are merged (n = 5)
Figure 1 shows P1 in 44% of Gen1 speakers and 64% of Gen2 speakers. The remaining Gen1 and Homeland (Gen0) speakers all showed P2. Meanwhile, P3 occurred exclusively among Gen2 speakers. Figures 2 and 3 show both lower Pillai Scores (more /y/~/u/ merger) and lower Cantonese Production Scores favoring P3 (suggesting English influence of UW on /y/), while higher scores favor P1 and P2 (suggesting Cantonese influence of /u/ on UW).
To conclude, by including a cross-linguistic comparison based on individual speaker production, these results highlight both contact and lack of contact as characteristics of bilingual phonetic variation in a North American immigrant context. Thus, in Gen1 (Hong Kong raised and generally Cantonese dominant) we find both P2 (Cantonese to English influence) and P1 (lack of contact). Similarly, in Gen2 (Toronto raised and generally English dominant) we find both P3 (English to Cantonese influence) and P1. Thus, while Gen2 shows more English influence as a group, Gen2 speakers are also more likely to produce both language internal and cross-linguistic distinctions.
Keywords
- bilingualism,
- heritage languages,
- Chinese - Yue,
- sound change,
- Cantonese,
- vowels,
- variation and change,
- sociophonetics,
- language contact
Disciplines
Publication Date
November 11, 2022
Location
University of Wisconsin - Madison, Madison, WI
Citation Information
Holman Tse. "Cross-linguistic comparisons of high vowels within individual speakers of Toronto Heritage Cantonese"" Workshop on Immigrant Languages in the Americas (WILA 13) (2022) Available at: http://works.bepress.com/holman-tse/19/
Creative Commons License
This work is licensed under a Creative Commons CC_BY International License.