Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
To trace how American English becomes the entrenched default in global AI systems, the authors audit structural dialectal bias across data curation, tokenization, representation, and generation using a benchmark of 1,813 matched American and British English variants. American English is systematically privileged across all six audited pretraining corpora, 21 post-training datasets, tokenizer compression ratios, and prediction costs, remaining the generative default even under neutral prompting. These results provide the first end-to-end empirical audit showing that dialectal homogenization is structurally baked into every stage of the LLM pipeline rather than emerging purely from inference-time behavior.
Even explicit non-American prompting fails to dislodge the American English default that is structurally hardcoded across pretraining data, tokenizers, and prediction costs in modern LLMs.
Large language models (LLMs) are increasingly embedded in educational, professional, and public infrastructure, yet widely used platforms expose "English (US)" as a primary English setting despite the global diversity of English. We ask: How does "English (US)" become the default? We study this question as structural bias, examining how geopolitical histories of data curation, digital dominance, and linguistic standardization intersect with the LLM development pipeline. Using British English as a controlled reference, we construct a curated resource of 1,813 matched American English (AmE)--British English (BrE) variants and introduce DiAlign, a dynamic, training-free method for estimating regional alignment from distributional evidence. We triangulate the AmE preference across data exposure --> representation --> generation, jointly examining pretraining and post-training data, tokenizer behavior and provenance, model prediction cost, and generated language across developer countries, prompt conditions, domains and sources, linguistic categories, and registers. AmE is consistently favored across all six audited pretraining corpora and 21 post-training datasets, is generally represented more compactly by tokenizers, and receives lower prediction cost. It also remains the dominant generation default under neutral English prompting; British-English prompting shifts this preference toward BrE but does not consistently eliminate the AmE default. To our knowledge, this is the first rigorous pipeline-wide study of structural bias across major phases of LLM development. Our findings show that contemporary LLMs privilege AmE as the de facto norm, raising concerns about linguistic homogenization, epistemic injustice, and inequity in global AI deployment, while providing a rigorous basis for targeted component-level intervention.