This dissertation investigates the roles of local identity, language attitude, social awareness, as well as social meanings in dialect change by examining Tianjin Chinese tone sandhi in apparent time. It studies the process by which local variants become stigmatized and also the process by which local features increase. Tianjin Chinese is in the process of standardization (Gao & Lu, 2003; Gu & Liu, 2003), but the current study finds that only stigmatized local features are disappearing, while an unmarked local feature seems to be immune to standardization. I interpret this in line with Labov's (1972) study of Martha's Vineyard, whereby traditional local features may come to index resistance to standardization and to the incursion of new people into the speech community. Ninety sociolinguistic interviews, including a word list, were conducted in Tianjin in the local dialect in 2014-16 (48f, age 18-82). Participants were categorized as 'middle class' or 'working class' using a combined measure of occupation, education and income. Qualitative assessments of 'positive' or 'negative' were assigned to speakers' attitudes to Tianjin and to migrants. The variables of the dissertation are three of the four traditional Tianjin tone sandhi, referred to as (53-53), (53-21) and (21-21) after their input tonal values respectively. Application of the (53-53) and (53-21) variables produce local outputs; non-application makes a speaker sound more like a Standard Chinese speaker. The old output variant of (21-21) is traditional; the new variant is closer to Standard Chinese. 7462 tokens of (53-53), 5683 tokens of (53-21), and 4117 tokens of (21-21) were extracted from the interviews and word lists (N = 17262). Tokens were impressionistically coded for the application or non-application of (53-53) and (53-21), and for the new or old variant of (21-21), and a subsample were checked in Praat. (53-53) application and the old variant of (21-21) have decreased substantially in frequency over time, probably because their outputs include or are similar to the most salient Tianjin low tone: Tone 1 (Han 1993). Mixed effects regression shows that they are avoided in word list style, by the middle class, by women, and by people with a negative attitude to Tianjin.In contrast, (53-21) has increased from 73.5% among speakers aged 65+, to 93% among speakers under 65. I speculate that because (53-21) is below public awareness, with little style-shifting, it is available for 'recycling' (Dubois & Horvath 2000) as a positive marker of 'new' Tianjin identity. Tianjin natives may be linguistically contrasting themselves with the many migrants who have moved to the city in the last three decades, and indeed a negative attitude to migrants significantly increases the likelihood of (53-21) application in the regress.The reduction or elimination of prominent stereotyped dialect features has been observed in languages with large numbers of speakers, often because of migration and language contacts (Hinskens, 1998). Unmarked dialect features have been observed to persist or increase to keep local identity (Labov 1972, Dubois & Horvath 2000). These two conflicting forces might lead to a stable compromise dialect (Hinskens, 1998). Here the case study of Tianjin Chinese tone sandhi also exhibits signs of changing to a compromise dialect, with stereotyped local features disappearing and unmarked local features increasing, adding to the expanding number of non-Western case studies of language change (Stanford & Preston, 2009) that support earlier generalizations made from Western communities.
Read
