Dataset description
This dataset contains replication data for the study "Performance context and genre conventions: The stability of African American English features in live and studio blues recordings" (De Timmerman, in review). The study investigates the stability of African American English (AAE) features in blues lyrics across different performance contexts. Using a corpus of 540 songs, each performed both live and in the studio by 45 artists from diverse sociocultural backgrounds and time periods, the study annotates eight phonological and lexico‑grammatical AAE features in 29,773 tokens. Through descriptive statistics and gradient‑boosted decision tree modeling with SHAP explainability, the results show that AAE features are consistently employed in blues performance, with minimal systematic differences between live and studio recordings, song types, time periods, or artist backgrounds. Instead, AAE realization is primarily driven by linguistic factors such as segmental context and lexical frequency. The findings support the argument that AAE features function as robust stylistic conventions in the blues, viz. as indexical of artistic authenticity, and increasingly iconic of the genre itself.
The dataset contains two data files: aae_in_blues.csv and aae_in_blues_metadata.csv. The aae_in_blues.csv file holds one row per AAE feature token (29,773 tokens in total), with each row recording the target word, its immediate lexical and phonological environment, whether the feature was realized, and the sociostylistic metadata of the recording in which it occurs (artist, song, performance context, song type, time period, and social group). The aae_in_blues_metadata.csv file holds metadata for all of the sampled recordings, including the link to the original recording on YouTube, the upload and access dates, and the year in which the original recording was made. All code used for the project is available at https://github.com/romeodetimmerman/aae-in-blues
(2026-09-29)
Article abstract
Following Trudgill’s (1983) seminal work on the linguistic tendencies of British singers, sociolinguists have shown that music provides a rich site for examining linguistic variation in performative contexts. While some scholarship on language use in music has touched on potential linguistic differences between live and studio performances (e.g. Coupland, 2011), it seems very few scholars have systematically researched this from a corpus-based perspective. Addressing this gap, the present study investigates whether the use of African American English (AAE) features in blues music varies across performance context: viz. live versus studio performance. Drawing on a corpus of 540 recordings of 270 blues songs, each performed both live and in the studio by 45 artists, the study quantitatively analyzes the frequency of five phonological and three lexico-grammatical features of AAE. Although we hypothesize differences in AAE feature use between live and studio recordings, descriptive statistical analysis and gradient boosted decision tree modeling reveal a high degree of consistency across both performance contexts. Moreover, SHAP interaction values show that the effect of performance context is hardly conditioned by song type or performers' socio-cultural background. This pattern mirrors previous findings showing that blues performers consistently employ AAE features regardless of their socio-cultural background (De Timmerman et al., 2023), or whether they are performing original songs or covers (De Timmerman and Slembrouck, 2024). Instead, AAE realization is primarily conditioned by linguistic factors, most notably the lexical frequency of the target word, the feature under analysis, and the place of articulation of the following segment. Our findings hence reinforce the argument that AAE features function as stable genre conventions in blues performance, whose indexical associations with the genre's origins (Silverstein, 2003) have become iconized as markers of authentic blues (Irvine and Gal, 2000), while remaining systematically conditioned by local linguistic context. (2026-09-29) |
