A Sociolinguistic Study of P-D-Drop in British English from the BNC1994 to BNC2014
(Accepted at English Language and Linguistics) This project investigates the sociolinguistic patterning of P-D-Drop in spoken British English — a construction in which the preposition "to" and determiner is omitted after directional motion verbs like go, come, and been (e.g., I went gym or come London this weekend). Drawing on the demographically sampled spoken components of the British National Corpus 1994 and 2014 as cross-sectional snapshots separated by two decades, the study examines how this variable is socially stratified by gender, socioeconomic class, region, and age, and whether it shows evidence of change over real time. Using mixed-effects logistic regression, the analysis reveals that to-drop functions as a vernacular feature associated with male and working-class speakers across England, consistent with a covert prestige account, while a significant decline between the two time periods — observed across all social groups simultaneously — points to communal rather than generational change, likely reflecting broader processes of dialect levelling in contemporary British English. These regional, social, and diachronic findings are interpreted as evidence of a possible change from above, driven by deindustrialisation, the erosion of dense working-class communities, and the cultural delegitimisation of working-class identity in post-industrial Britain.
Acquisition of Classifiers in Heritage Mandarin: Parental Input Quality, Register Exposure, and Developmental Metrics Beyond Age
(Under Review) Heritage speakers are often characterized through a deficit narrative, with their grammatical differences partly attributed to the quality of parental input. This study challenges this assumption by investigating production of Mandarin nominal count classifiers—a domain vulnerable in heritage grammars that is also conditioned by discourse factors—using CHILDES data from four mother–child dyads of monolingual (China) and heritage (Chinese-American) backgrounds. Mixed-effect logistic regressions show that while heritage children substitute the general classifier more often than monolinguals, both groups produce grammatically coherent systems with negligible syntactic errors. Indicators of child linguistic stage, Mean Length of Utterance and noun diversity, predict classifier substitution in children and parents better than age, with both the monolingual and heritage group showing parallel trajectories. This underscores the importance of adopting linguistically informed developmental metrics in heritage language research, given the greater age-related variability. Crucially, mothers of heritage and monolingual children do not differ significantly in their classifier input quality regarding substitution. These findings suggest that differential outcomes in heritage speakers reflect reduced input quantity and limited register exposure, not deficient parental input quality.
The Socioemotional Risk of Heritage Language Loss in Childhood: A U.S. Nationally Representative Longitudinal Analysis
(In Preparation, in collaboration with Susanne Gahl) Heritage language loss among immigrant children is rapid and widespread, yet its psychological consequences remain underexamined in large-scale longitudinal research. This study uses the Early Childhood Longitudinal Study, Kindergarten Class of 2010–11 (ECLS-K:2011) — a nationally representative sample — to construct heritage language use trajectories from kindergarten through fifth grade and examine their relationship to internalizing problem behaviors. Five trajectory groups are identified: Highly maintained, Improved, Moderately maintained, Diminished, and Complete Loss. In a weighted regression model adjusting for demographic and social covariates, children who completely lost their heritage language by fifth grade showed significantly higher teacher-rated internalizing symptoms — including anxiety, loneliness, low self-esteem, and sadness — compared to children who maintained use. These findings hold even after accounting for children's satisfaction with parental attention, pointing to a unique developmental role of the heritage language beyond general family warmth. The results support a biculturalism framework in which heritage language maintenance fosters ethnic identity, communicative intimacy, and psychological security, and advocate for additive bilingual programs as a matter of children's socioemotional health.
The Lexical and Syntactic Input–Output Relationship in Heritage Language Acquisition: Evidence from Spanish Bilingual and Monolingual Parent-Child Dyads
(In Preparation) Heritage speakers' grammatical differences from monolingual norms are often attributed to deficient parental input, yet few studies directly examine input quality using naturalistic data. This study investigates whether bilingual mothers' child-directed Spanish differs from that of monolingual mothers in syntactic complexity and lexical diversity, and how maternal speech relates to children's own production. Drawing on parent–child dyadic data from the CHILDES Hoff and Fernández/Aguado corpora at ages 3;0 and 3;6, it applies mean length of utterance and word type counts to both speakers within each dyad, models concurrent and lagged input–output relationships, and examines how family-level environmental factors predict language outcomes in the bilingual group. Contrary to deficit expectations, bilingual mothers produced significantly more lexical types than monolingual mothers across time points and word classes—yet bilingual children still exhibited lower lexical diversity than their monolingual peers. The proportion of Spanish input heard at home was the strongest predictor of bilingual children's MLU and lexical diversity, outperforming any measure of input quality. These findings suggest that heritage children's more limited Spanish production reflects reduced quantity and register diversity of heritage language exposure, rather than impoverished input quality from their caregivers.