Investigating African-American Vernacular English in Transformer-Based Text Generation
This addresses the problem of NLP model bias for underrepresented language varieties like AAVE, which is important for social media users and AI fairness, but it is incremental as it builds on existing methods with new data.
The study tackled the performance of GPT-2 on African American Vernacular English (AAVE) by creating a parallel dataset of AAVE and Standard American English tweets, finding that AAVE text leads to more negative sentiment classifications but GPT-2 increases positive sentiment for both, with human evaluations comparing contextual rigor and quality.
The growth of social media has encouraged the written use of African American Vernacular English (AAVE), which has traditionally been used only in oral contexts. However, NLP models have historically been developed using dominant English varieties, such as Standard American English (SAE), due to text corpora availability. We investigate the performance of GPT-2 on AAVE text by creating a dataset of intent-equivalent parallel AAVE/SAE tweet pairs, thereby isolating syntactic structure and AAVE- or SAE-specific language for each pair. We evaluate each sample and its GPT-2 generated text with pretrained sentiment classifiers and find that while AAVE text results in more classifications of negative sentiment than SAE, the use of GPT-2 generally increases occurrences of positive sentiment for both. Additionally, we conduct human evaluation of AAVE and SAE text generated with GPT-2 to compare contextual rigor and overall quality.